Lu Yin

Lu Yin, You Wu, Zhenyu Zhang et al.

Large Language Models (LLMs), renowned for their remarkable performance across diverse domains, present a challenge when it comes to practical deployment due to their colossal model size. In response to this challenge, efforts have been directed toward the application of traditional network pruning techniques to LLMs, uncovering a massive number of parameters that can be pruned in one-shot without hurting performance. Prevailing LLM pruning strategies have consistently adhered to the practice of uniformly pruning all layers at equivalent sparsity, resulting in robust performance. However, this observation stands in contrast to the prevailing trends observed in the field of vision models, where non-uniform layerwise sparsity typically yields stronger results. To understand the underlying reasons for this disparity, we conduct a comprehensive study and discover a strong correlation with the emergence of activation outliers in LLMs. Inspired by this finding, we introduce a novel LLM pruning methodology that incorporates a tailored set of non-uniform layerwise sparsity ratios, termed as Outlier Weighed Layerwise sparsity (OWL). The sparsity ratio of OWL is proportional to the outlier ratio observed within each layer, facilitating a more effective alignment between layerwise weight sparsity and outlier ratios. Our empirical evaluation, conducted across the LLaMA-V1 family and OPT, spanning various benchmarks, demonstrates the distinct advantages offered by OWL over previous methods. For instance, OWL exhibits a remarkable performance gain, surpassing the state-of-the-art Wanda and SparseGPT by 61.22 and 6.80 perplexity at a high sparsity level of 70%, respectively, while delivering 2.6x end-to-end inference speed-up in the DeepSparse inference engine. Codes are available at https://github.com/luuyin/OWL.

Zahra Atashgahi, Xuhao Zhang, Neil Kichler et al.

Feature selection that selects an informative subset of variables from data not only enhances the model interpretability and performance but also alleviates the resource demands. Recently, there has been growing attention on feature selection using neural networks. However, existing methods usually suffer from high computational costs when applied to high-dimensional datasets. In this paper, inspired by evolution processes, we propose a novel resource-efficient supervised feature selection method using sparse neural networks, named \enquote{NeuroFS}. By gradually pruning the uninformative features from the input layer of a sparse neural network trained from scratch, NeuroFS derives an informative subset of features efficiently. By performing several experiments on $11$ low and high-dimensional real-world benchmarks of different types, we demonstrate that NeuroFS achieves the highest ranking-based score among the considered state-of-the-art supervised feature selection models. The code is available on GitHub.

Ajay Jaiswal, Yifan Wang, Lu Yin et al.

Large Language Models' (LLMs) weight matrices can often be expressed in low-rank form with potential to relax memory and compute resource requirements. Unlike prior efforts that focus on developing novel matrix decompositions, in this work we study the non-uniform low-rank properties of weight matrices in LLMs through the lens of stabilizing gradient subspace. First, we provide a theoretical framework to understand the stabilization of gradient subspaces through Hessian analysis. Second, we empirically establish an important relationship between gradient dynamics and low-rank expressiveness of weight matrices. Our findings reveal that different LLM components exhibit varying levels of converged low-rank structures, necessitating variable rank reduction across them to minimize drop in performance due to compression. Drawing on this result, we present Weight Low-Rank Projection(WeLore) that unifies weight compression and memory-efficient fine-tuning into one, in a data-agnostic and one-shot manner. When used as a compression technique, WeLore categorizes weight matrices into Low-rank Components (LRCs) and Non-Low-rank Components (N-LRCs) and suitably encodes them for minimum performance loss. Our gradient dynamics perspective illustrates that LRCs tend to have better fine-tuning capabilities and their standalone fine-tuning can closely mimic and sometimes outperform the training loss trajectory and performance of full fine-tuning with notable memory and compute footprint reduction. Codes are available at https://github.com/VITA-Group/WeLore.

Qiao Xiao, Boqian Wu, Lu Yin et al.

While deep learning has demonstrated impressive progress, it remains a daunting challenge to learn from hard samples as these samples are usually noisy and intricate. These hard samples play a crucial role in the optimal performance of deep neural networks. Most research on Sparse Neural Networks (SNNs) has focused on standard training data, leaving gaps in understanding their effectiveness on complex and challenging data. This paper's extensive investigation across scenarios reveals that most SNNs trained on challenging samples can often match or surpass dense models in accuracy at certain sparsity levels, especially with limited data. We observe that layer-wise density ratios tend to play an important role in SNN performance, particularly for methods that train from scratch without pre-trained initialization. These insights enhance our understanding of SNNs' behavior and potential for efficient learning approaches in data-centric AI. Our code is publicly available at: \url{https://github.com/QiaoXiao7282/hard_sample_learners}.

Lu Yin, Ajay Jaiswal, Shiwei Liu et al.

We present Junk DNA Hypothesis by adopting a novel task-centric angle for the pre-trained weights of large language models (LLMs). It has been believed that weights in LLMs contain significant redundancy, leading to the conception that a considerable chunk of the parameters can be removed by pruning without compromising performance. Contrary to this belief, this paper presents a counter-argument: small-magnitude weights of pre-trained model weights encode vital knowledge essential for tackling difficult downstream tasks - manifested as the monotonic relationship between the performance drop of downstream tasks across the difficulty spectrum, as we prune more pre-trained weights by magnitude. Moreover, we reveal that these seemingly inconsequential weights can result in irreparable loss of knowledge and performance degradation in difficult tasks, even when downstream continual training is allowed. Interestingly, our evaluations show that the other popular compression, namely quantization, fails to exhibit similar monotonic effect and does not as convincingly disentangle this task-difficulty information. To study formally, we introduce several quantifiable metrics to gauge the downstream task difficulty: (1) within the same task category, and (2) across different task categories. Our extensive experiments substantiate the Junk DNA Hypothesis across a diverse range of model sizes, tasks, datasets, and even pruning methods. Codes are available at: https://github.com/VITA-Group/Junk_DNA_Hypothesis.git.

Tianjin Huang, Tianlong Chen, Meng Fang et al.

Recent works have impressively demonstrated that there exists a subnetwork in randomly initialized convolutional neural networks (CNNs) that can match the performance of the fully trained dense networks at initialization, without any optimization of the weights of the network (i.e., untrained networks). However, the presence of such untrained subnetworks in graph neural networks (GNNs) still remains mysterious. In this paper we carry out the first-of-its-kind exploration of discovering matching untrained GNNs. With sparsity as the core tool, we can find \textit{untrained sparse subnetworks} at the initialization, that can match the performance of \textit{fully trained dense} GNNs. Besides this already encouraging finding of comparable performance, we show that the found untrained subnetworks can substantially mitigate the GNN over-smoothing problem, hence becoming a powerful tool to enable deeper GNNs without bells and whistles. We also observe that such sparse untrained subnetworks have appealing performance in out-of-distribution detection and robustness of input perturbations. We evaluate our method across widely-used GNN architectures on various popular datasets including the Open Graph Benchmark (OGB).

Di He, Songjun Tu, Keyu Wang et al.

Learning rate configuration is a fundamental aspect of modern deep learning. The prevailing practice of applying a uniform learning rate across all layers overlooks the structural heterogeneity of Transformers, potentially limiting their effectiveness as the backbone of Large Language Models (LLMs). In this paper, we introduce Layerwise Learning Rate (LLR), an adaptive scheme that assigns distinct learning rates to individual Transformer layers. Our method is grounded in Heavy-Tailed Self-Regularization (HT-SR) theory, which characterizes the empirical spectral density (ESD) of weight correlation matrices to quantify heavy-tailedness. Layers with weaker heavy-tailedness are assigned larger learning rates to accelerate their training, while layers with stronger heavy-tailedness receive smaller learning rates. By tailoring learning rates in this manner, LLR promotes balanced training across layers, leading to faster convergence and improved generalization. Extensive experiments across architectures (from LLaMA to GPT-nano), optimizers (AdamW and Muon), and parameter scales (60M-1B) demonstrate that LLR achieves up to 1.5x training speedup and outperforms baselines, notably raising average zero-shot accuracy from 47.09% to 49.02%. A key advantage of LLR is its low tuning overhead: it transfers nearly optimal LR settings directly from the uniform baseline. Code is available at https://github.com/hed-ucas/Layer-wise-Learning-Rate.

Xiaolong Han, Zehong Wang, Bo Zhao et al.

Neural network weights are typically viewed as the end product of training, while most deep learning research focuses on data, features, and architectures. However, recent advances show that the set of all possible weight values (weight space) itself contains rich structure: pretrained models form organized distributions, exhibit symmetries, and can be embedded, compared, or even generated. Understanding such structures has tremendous impact on how neural networks are analyzed and compared, and on how knowledge is transferred across models, beyond individual training instances. This emerging research direction, which we refer to as Weight Space Learning (WSL), treats neural weights as a meaningful domain for analysis and modeling. This survey provides the first unified taxonomy of WSL. We categorize existing methods into three core dimensions: Weight Space Understanding (WSU), which studies the geometry and symmetries of weights; Weight Space Representation (WSR), which learns embeddings over model weights; and Weight Space Generation (WSG), which synthesizes new weights through hypernetworks or generative models. We further show how these developments enable practical applications, including model retrieval, continual and federated learning, neural architecture search, and data-free reconstruction. By consolidating fragmented progress under a coherent framework, this survey highlights weight space as a learnable, structured domain with growing impact across model analysis, transferring, and weight generation. We release an accompanying resource at https://github.com/Zehong-Wang/Awesome-Weight-Space-Learning.

Lu Yin, Shiwei Liu, Meng Fang et al.

Lottery tickets (LTs) is able to discover accurate and sparse subnetworks that could be trained in isolation to match the performance of dense networks. Ensemble, in parallel, is one of the oldest time-proven tricks in machine learning to improve performance by combining the output of multiple independent models. However, the benefits of ensemble in the context of LTs will be diluted since ensemble does not directly lead to stronger sparse subnetworks, but leverages their predictions for a better decision. In this work, we first observe that directly averaging the weights of the adjacent learned subnetworks significantly boosts the performance of LTs. Encouraged by this observation, we further propose an alternative way to perform an 'ensemble' over the subnetworks identified by iterative magnitude pruning via a simple interpolating strategy. We call our method Lottery Pools. In contrast to the naive ensemble which brings no performance gains to each single subnetwork, Lottery Pools yields much stronger sparse subnetworks than the original LTs without requiring any extra training or inference cost. Across various modern architectures on CIFAR-10/100 and ImageNet, we show that our method achieves significant performance gains in both, in-distribution and out-of-distribution scenarios. Impressively, evaluated with VGG-16 and ResNet-18, the produced sparse subnetworks outperform the original LTs by up to 1.88% on CIFAR-100 and 2.36% on CIFAR-100-C; the resulting dense network surpasses the pre-trained dense-model up to 2.22% on CIFAR-100 and 2.38% on CIFAR-100-C.

Lu Yin, Vlado Menkovski, Meng Fang et al.

Recent works on sparse neural network training (sparse training) have shown that a compelling trade-off between performance and efficiency can be achieved by training intrinsically sparse neural networks from scratch. Existing sparse training methods usually strive to find the best sparse subnetwork possible in one single run, without involving any expensive dense or pre-training steps. For instance, dynamic sparse training (DST), is capable of reaching a competitive performance of dense training by iteratively evolving the sparse topology during the course of training. In this paper, we argue that it is better to allocate the limited resources to create multiple low-loss sparse subnetworks and superpose them into a stronger one, instead of allocating all resources entirely to find an individual subnetwork. To achieve this, two desiderata are required: (1) efficiently producing many low-loss subnetworks, the so-called cheap tickets, within one training process limited to the standard training time used in dense training; (2) effectively superposing these cheap tickets into one stronger subnetwork. To corroborate our conjecture, we present a novel sparse training approach, termed Sup-tickets, which can satisfy the above two desiderata concurrently in a single sparse-to-sparse training process. Across various modern architectures on CIFAR-10/100 and ImageNet, we show that Sup-tickets integrates seamlessly with the existing sparse training methods and demonstrates consistent performance improvement.

Pengxiang Li, Dilxat Muhtar, Tianlong Chen et al.

Diffusion Language Models (DLMs) are often advertised as enabling parallel token generation, yet practical fast DLMs frequently converge to left-to-right, autoregressive (AR)-like decoding dynamics. In contrast, genuinely non-AR generation is promising because it removes AR's sequential bottleneck, better exploiting parallel hardware to reduce synchronization/communication overhead and improve latency scaling with output length. We argue that a primary driver of AR-like decoding is a mismatch between DLM objectives and the highly sequential structure of widely used training data, including standard pretraining corpora and long chain-of-thought (CoT) supervision. Motivated by this diagnosis, we propose NAP (Non-Autoregressive Parallel DLMs), a proof-of-concept, data-centric approach that better aligns supervision with non-AR parallel decoding. NAP curates examples as multiple independent reasoning trajectories and couples them with a parallel-forced decoding strategy that encourages multi-token parallel updates. Across math reasoning benchmarks, NAP yields stronger performance under parallel decoding than DLMs trained on standard long CoT data, with gains growing as parallelism increases. Our results suggest that revisiting data and supervision is a principled direction for mitigating AR-like behavior and moving toward genuinely non-autoregressive parallel generation in DLMs. Our code is available at https://github.com/pixeli99/NAP.

Xiaolong Han, Ferrante Neri, Zijian Jiang et al.

Each LoRA checkpoint compactly stores task-specific updates in low-rank weight matrices, offering an efficient way to adapt large language models to new tasks and domains. In principle, these weights already encode what the adapter does and how well it performs. In this paper, we ask whether this information can be read directly from the weights, without running the base model or accessing training data. A key obstacle is that a single LoRA update can be factorized in infinitely many ways. Without resolving this ambiguity, models trained on the factors may fit the particular factorization rather than the underlying update. To this end, we propose \methodfull, which maps each LoRA update to a provably canonical form via QR decomposition followed by SVD, so that all equivalent factorizations share the same representation. The resulting components are then tokenized and processed by a Transformer to produce a weight-space embedding. Across language and vision LoRA collections, W2T achieves strong results on attribute classification, performance prediction, and adapter retrieval, demonstrating that LoRA weights reliably indicate model behavior once factorization ambiguity is removed. Code is available at https://github.com/xiaolonghan2000/Weight2Token.

Zhenyu Zhang, Ajay Jaiswal, Lu Yin et al.

Training Large Language Models (LLMs) is memory-intensive due to the large number of parameters and associated optimization states. GaLore, a recent method, reduces memory usage by projecting weight gradients into a low-rank subspace without compromising performance. However, GaLore relies on time-consuming Singular Value Decomposition (SVD) operations to identify the subspace, and the frequent subspace updates lead to significant training time overhead. Moreover, GaLore offers minimal improvements in accuracy and efficiency compared to LoRA in more accessible fine-tuning scenarios. To address these limitations, we introduce Q-Galore, a novel approach that substantially reduces memory usage by combining quantization and low-rank projection, surpassing the benefits of GaLore. Our method is based on two key observations: (i) the gradient subspace exhibits diverse properties, with some layers converging early in training while others are subject to frequent changes; (ii) the projection matrices are highly resilient to low-bit quantization. Leveraging these insights, Q-GaLore adaptively updates the gradient subspace based on its convergence statistics, achieving comparable performance while significantly reducing the number of SVD operations. We maintain the projection matrices in INT4 format and weights in INT8 format, incorporating stochastic rounding to capture accumulated gradient information. This approach enables a high-precision training trajectory using only low-precision weights. We demonstrate that Q-GaLore achieves highly competitive performance with exceptional memory efficiency. At pre-training, Q-GaLore facilitates training a LLaMA-7B model from scratch on a single NVIDIA RTX 4060 Ti with only 16 GB memory. At fine-tuning, it reduces memory consumption by up to 50% compared to LoRA and GaLore, while consistently outperforming QLoRA at the same memory cost.

Tianjin Huang, Shiwei Liu, Tianlong Chen et al.

Despite the fact that adversarial training has become the de facto method for improving the robustness of deep neural networks, it is well-known that vanilla adversarial training suffers from daunting robust overfitting, resulting in unsatisfactory robust generalization. A number of approaches have been proposed to address these drawbacks such as extra regularization, adversarial weights perturbation, and training with more data over the last few years. However, the robust generalization improvement is yet far from satisfactory. In this paper, we approach this challenge with a brand new perspective -- refining historical optimization trajectories. We propose a new method named \textbf{Weighted Optimization Trajectories (WOT)} that leverages the optimization trajectories of adversarial training in time. We have conducted extensive experiments to demonstrate the effectiveness of WOT under various state-of-the-art adversarial attacks. Our results show that WOT integrates seamlessly with the existing adversarial training methods and consistently overcomes the robust overfitting issue, resulting in better adversarial robustness. For example, WOT boosts the robust accuracy of AT-PGD under AA-$L_{\infty}$ attack by 1.53\% $\sim$ 6.11\% and meanwhile increases the clean accuracy by 0.55\%$\sim$5.47\% across SVHN, CIFAR-10, CIFAR-100, and Tiny-ImageNet datasets.

Ricky Maulana Fajri, Yulong Pei, Lu Yin et al.

Despite significant advancements in active learning and adversarial attacks, the intersection of these two fields remains underexplored, particularly in developing robust active learning frameworks against dynamic adversarial threats. The challenge of developing robust active learning frameworks under dynamic adversarial attacks is critical, as these attacks can lead to catastrophic forgetting within the active learning cycle. This paper introduces Robust Active Learning (RoAL), a novel approach designed to address this issue by integrating Elastic Weight Consolidation (EWC) into the active learning process. Our contributions are threefold: First, we propose a new dynamic adversarial attack that poses significant threats to active learning frameworks. Second, we introduce a novel method that combines EWC with active learning to mitigate catastrophic forgetting caused by dynamic adversarial attacks. Finally, we conduct extensive experimental evaluations to demonstrate the efficacy of our approach. The results show that RoAL not only effectively counters dynamic adversarial threats but also significantly reduces the impact of catastrophic forgetting, thereby enhancing the robustness and performance of active learning systems in adversarial environments.

Jiaxi Li, Lu Yin, Li Shen et al.

Large Language Models (LLMs) have achieved remarkable capabilities, but their immense computational demands during training remain a critical bottleneck for widespread adoption. Low-rank training has received attention in recent years due to its ability to significantly reduce training memory usage. Meanwhile, applying 2:4 structured sparsity to weights and activations to leverage NVIDIA GPU support for 2:4 structured sparse format has become a promising direction. However, existing low-rank methods often leave activation matrices in full-rank, which dominates memory consumption and limits throughput during large-batch training. Furthermore, directly applying sparsity to weights often leads to non-negligible performance degradation. To achieve efficient pre-training of LLMs, this paper proposes ELAS: Efficient pre-training of Low-rank LLMs via 2:4 Activation Sparsity, a novel framework for low-rank models via 2:4 activation sparsity. ELAS applies squared ReLU activation functions to the feed-forward networks in low-rank models and implements 2:4 structured sparsity on the activations after the squared ReLU operation. We evaluated ELAS through pre-training experiments on LLaMA models ranging from 60M to 1B parameters. The results demonstrate that ELAS maintains performance with minimal degradation after applying 2:4 activation sparsity, while achieving training and inference acceleration. Moreover, ELAS reduces activation memory overhead, particularly with large batch sizes. Code is available at ELAS Repo.

Xiangxiang Cui, Tianjin Huang, Yifang Wang et al.

Medical foundation models (MedFMs) have emerged as transformative tools in healthcare, demonstrating capabilities across diverse clinical applications. These models can be broadly categorized into two paradigms: Medical Vision-Language Models (Med-VLMs) and segmentation foundation models. Med-VLMs range from medical-specialized models such as LLaVA-Med and MedGemma, to general-purpose models like GPT-4o and Gemini, all capable of medical image understanding tasks including visual question answering (VQA), report generation, and visual grounding. Concurrently, the Segment Anything Model (SAM) has catalyzed a new generation of medical segmentation models, with adaptations like SAM-Med2D and MedSAM. The widespread clinical deployment of these models thus necessitates rigorous evaluation of their reliability under real-world conditions.

Pengxiang Li, Lu Yin, Shiwei Liu

Large Language Models (LLMs) have achieved remarkable success, yet recent findings reveal that their deeper layers often contribute minimally and can be pruned without affecting overall performance. While some view this as an opportunity for model compression, we identify it as a training shortfall rooted in the widespread use of Pre-Layer Normalization (Pre-LN). We demonstrate that Pre-LN, commonly employed in models like GPT and LLaMA, leads to diminished gradient norms in its deeper layers, reducing their effectiveness. In contrast, Post-Layer Normalization (Post-LN) preserves larger gradient norms in deeper layers but suffers from vanishing gradients in earlier layers. To address this, we introduce Mix-LN, a novel normalization technique that combines the strengths of Pre-LN and Post-LN within the same model. Mix-LN applies Post-LN to the earlier layers and Pre-LN to the deeper layers, ensuring more uniform gradients across layers. This allows all parts of the network--both shallow and deep layers--to contribute effectively to training. Extensive experiments with various model sizes from 70M to 7B demonstrate that Mix-LN consistently outperforms both Pre-LN and Post-LN, promoting more balanced, healthier gradient norms throughout the network, and enhancing the overall quality of LLM pre-training. Furthermore, we demonstrate that models pre-trained with Mix-LN learn better compared to those using Pre-LN or Post-LN during supervised fine-tuning (SFT) and reinforcement learning from human feedback (RLHF), highlighting the critical importance of high-quality deep layers. By effectively addressing the inefficiencies of deep layers in current LLMs, Mix-LN unlocks their potential, enhancing model capacity without increasing model size. Our code is available at https://github.com/pixeli99/MixLN.

Wenfang Sun, Xinyuan Song, Pengxiang Li et al.

In this paper, we introduce the Curse of Depth, a concept that highlights, explains, and addresses the recent observation in modern Large Language Models (LLMs) where nearly half of the layers are less effective than expected. We first confirm the wide existence of this phenomenon across the most popular families of LLMs such as Llama, Mistral, DeepSeek, and Qwen. Our analysis, theoretically and empirically, identifies that the underlying reason for the ineffectiveness of deep layers in LLMs is the widespread usage of Pre-Layer Normalization (Pre-LN). While Pre-LN stabilizes the training of Transformer LLMs, its output variance exponentially grows with the model depth, which undesirably causes the derivative of the deep Transformer blocks to be an identity matrix, and therefore barely contributes to the training. To resolve this training pitfall, we propose LayerNorm Scaling (LNS), which scales the variance of output of the layer normalization inversely by the square root of its depth. This simple modification mitigates the output variance explosion of deeper Transformer layers, improving their contribution. Across a wide range of model sizes (130M to 7B), our experiments show that LNS consistently outperforms previous normalization and scaling techniques in enhancing LLM pre-training performance. Moreover, this improvement seamlessly carries over to supervised fine-tuning. All these gains can be attributed to the fact that LayerNorm Scaling enables deeper layers to contribute more effectively during training. Our code is available at \href{https://github.com/lmsdss/LayerNorm-Scaling}{LayerNorm-Scaling}.

Pengxiang Li, Yefan Zhou, Dilxat Muhtar et al.

Diffusion language models (DLMs) have recently emerged as an alternative to autoregressive approaches, offering parallel sequence generation and flexible token orders. However, their inference remains slower than that of autoregressive models, primarily due to the cost of bidirectional attention and the large number of refinement steps required for high quality outputs. In this work, we highlight and leverage an overlooked property of DLMs early answer convergence: in many cases, the correct answer can be internally identified by half steps before the final decoding step, both under semi-autoregressive and random remasking schedules. For example, on GSM8K and MMLU, up to 97% and 99% of instances, respectively, can be decoded correctly using only half of the refinement steps. Building on this observation, we introduce Prophet, a training-free fast decoding paradigm that enables early commit decoding. Specifically, Prophet dynamically decides whether to continue refinement or to go "all-in" (i.e., decode all remaining tokens in one step), using the confidence gap between the top-2 prediction candidates as the criterion. It integrates seamlessly into existing DLM implementations, incurs negligible overhead, and requires no additional training. Empirical evaluations of LLaDA-8B and Dream-7B across multiple tasks show that Prophet reduces the number of decoding steps by up to 3.4x while preserving high generation quality. These results recast DLM decoding as a problem of when to stop sampling, and demonstrate that early decode convergence provides a simple yet powerful mechanism for accelerating DLM inference, complementary to existing speedup techniques. Our code is publicly available at https://github.com/pixeli99/Prophet.

Boqian Wu, Qiao Xiao, Shiwei Liu et al.

Deep neural networks have evolved as the leading approach in 3D medical image segmentation due to their outstanding performance. However, the ever-increasing model size and computation cost of deep neural networks have become the primary barrier to deploying them on real-world resource-limited hardware. In pursuit of improving performance and efficiency, we propose a 3D medical image segmentation model, named Efficient to Efficient Network (E2ENet), incorporating two parametrically and computationally efficient designs. i. Dynamic sparse feature fusion (DSFF) mechanism: it adaptively learns to fuse informative multi-scale features while reducing redundancy. ii. Restricted depth-shift in 3D convolution: it leverages the 3D spatial information while keeping the model and computational complexity as 2D-based methods. We conduct extensive experiments on BTCV, AMOS-CT and Brain Tumor Segmentation Challenge, demonstrating that E2ENet consistently achieves a superior trade-off between accuracy and efficiency than prior arts across various resource constraints. E2ENet achieves comparable accuracy on the large-scale challenge AMOS-CT, while saving over 68\% parameter count and 29\% FLOPs in the inference phase, compared with the previous best-performing method. Our code has been made available at: https://github.com/boqian333/E2ENet-Medical.

Xinchen Han, Hossam Afifi, Michel Marot et al.

Large Language Models (LLMs) often generate unnecessarily verbose Chain-of-Thought (CoT) reasoning that increases computational costs and latency without proportional performance gains. In this paper, we propose \textbf{F}ine-grained \textbf{G}roup policy \textbf{O}ptimization (\textbf{FGO}), a Reinforcement Learning (RL) algorithm that refines group responses by subdividing them and assigning appropriate weights based on length and entropy, thereby enabling effective CoT compression. Meanwhile, as an enhanced variant of Group Relative Policy Optimization (GRPO), FGO successfully addresses two major limitations of the GRPO: inefficient data utilization and entropy collapse. We evaluate FGO on multiple reasoning LLMs and benchmarks, including MATH500, AIME24, AMC23, and Minerva. Experimental results show that FGO achieves efficient CoT compression without degrading performance, and simultaneously resolves the key limitations of GRPO.

Zheyan Qu, Lu Yin, Zitong Yu et al.

Large language models (LLMs) have demonstrated astonishing capabilities in natural language processing (NLP) tasks, sparking interest in their application to professional domains with higher specialized requirements. However, restricted access to closed-source LLMs via APIs and the difficulty in collecting massive high-quality datasets pose obstacles to the development of large language models in education fields of various courses. Given these challenges, we propose CourseGPT-zh, a course-oriented education LLM that supports customization and low-cost deployment. To address the comprehensiveness and diversity requirements of course-specific corpora, we design a high-quality question-answering corpus distillation framework incorporating prompt optimization, which effectively mines textbook knowledge and enhances its diversity. Moreover, considering the alignment of LLM responses with user needs, a novel method for discrete prompt optimization based on LLM-as-Judge is introduced. During optimization, this framework leverages the LLM's ability to reflect on and exploit error feedback and patterns, allowing for prompts that meet user needs and preferences while saving response length. Lastly, we obtain CourseGPT-zh based on the open-source LLM using parameter-efficient fine-tuning. Experimental results show that our discrete prompt optimization framework effectively improves the response quality of ChatGPT, and CourseGPT-zh exhibits strong professional capabilities in specialized knowledge question-answering, significantly outperforming comparable open-source models.

Yuxiang Guo, Lu Yin, Bo Jiang et al.

Aligning large language models (LLMs) with human intent is critical for enhancing their performance across a variety of tasks. Standard alignment techniques, such as Direct Preference Optimization (DPO), often rely on the binary Bradley-Terry (BT) model, which can struggle to capture the complexities of human preferences -- particularly in the presence of noisy or inconsistent labels and frequent ties. To address these limitations, we introduce the Tie-rank Oriented Bradley-Terry model (TOBT), an extension of the BT model that explicitly incorporates ties, enabling more nuanced preference representation. Building on this, we propose Tie-rank Oriented Direct Preference Optimization (TODO), a novel alignment algorithm that leverages TOBT's ternary ranking system to improve preference alignment. In evaluations on Mistral-7B and Llama 3-8B models, TODO consistently outperforms DPO in modeling preferences across both in-distribution and out-of-distribution datasets. Additional assessments using MT Bench and benchmarks such as Piqa, ARC-c, and MMLU further demonstrate TODO's superior alignment performance. Notably, TODO also shows strong results in binary preference alignment, highlighting its versatility and potential for broader integration into LLM alignment. The implementation details can be found in https://github.com/XXares/TODO.

Jiaxu Zhao, Lu Yin, Shiwei Liu et al.

The deep neural network (DNN) has been proven effective in various domains. However, they often struggle to perform well on certain minority groups during inference, despite showing strong performance on the majority of data groups. This is because over-parameterized models learned \textit{bias attributes} from a large number of \textit{bias-aligned} training samples. These bias attributes are strongly spuriously correlated with the target variable, causing the models to be biased towards spurious correlations (i.e., \textit{bias-conflicting}). To tackle this issue, we propose a novel \textbf{re}weighted \textbf{s}parse \textbf{t}raining framework, dubbed as \textit{\textbf{REST}}, which aims to enhance the performance of biased data while improving computation and memory efficiency. Our proposed REST framework has been experimentally validated on three datasets, demonstrating its effectiveness in exploring unbiased subnetworks. We found that REST reduces the reliance on spuriously correlated features, leading to better performance across a wider range of data groups with fewer training and inference resources. We highlight that the \textit{REST} framework represents a promising approach for improving the performance of DNNs on biased data, while simultaneously improving computation and memory efficiency. By reducing the reliance on spurious correlations, REST has the potential to enhance the robustness of DNNs and improve their generalization capabilities. Code is released at \url{https://github.com/zhao1402072392/REST}

Zihang Liu, Tianyu Pang, Oleg Balabanov et al.

Recent studies have shown that supervised fine-tuning of LLMs on a small number of high-quality datasets can yield strong reasoning capabilities. However, full fine-tuning (Full FT), while powerful, is computationally expensive and susceptible to overfitting and catastrophic forgetting, particularly when data is limited. Sparse fine-tuning, which previously achieved notable success by updating only a small subset of model parameters, offers a promising trade-off between efficiency and effectiveness. Yet, it has lagged behind in the LLM era due to the difficulty of identifying parameters truly critical for reasoning. In this work, we state that weights with the largest magnitude after low-rank approximation are critical weights for fine-tuning, which we call Principal Weights. Surprisingly, while magnitude-based sparse fine-tuning performs poorly as a baseline on LLM fine-tuning, it becomes highly effective after rank reduction. These insights motivate our method: Low-rank Informed Sparse Fine-Tuning (LIFT). LIFT only updates the top 5% Principal Weights throughout training and consistently achieves better performance on reasoning tasks than Full FT, while maintaining memory efficiency on par with popular parameter-efficient fine-tuning methods. In addition to strong performance on target domains such as arithmetic reasoning, LIFT also retains up to 20% more source-domain knowledge, compared to Full FT and LoRA. Our code is available at: https://github.com/zihanghliu/LIFT.

Jinchuan Zhang, Lu Yin, Yan Zhou et al.

The acquisition of agentic capabilities has transformed LLMs from "knowledge providers" to "action executors", a trend that while expanding LLMs' capability boundaries, significantly increases their susceptibility to malicious use. Previous work has shown that current LLM-based agents execute numerous malicious tasks even without being attacked, indicating a deficiency in agentic use safety alignment during the post-training phase. To address this gap, we propose AgentAlign, a novel framework that leverages abstract behavior chains as a medium for safety alignment data synthesis. By instantiating these behavior chains in simulated environments with diverse tool instances, our framework enables the generation of highly authentic and executable instructions while capturing complex multi-step dynamics. The framework further ensures model utility by proportionally synthesizing benign instructions through non-malicious interpretations of behavior chains, precisely calibrating the boundary between helpfulness and harmlessness. Evaluation results on AgentHarm demonstrate that fine-tuning three families of open-source models using our method substantially improves their safety (35.8% to 79.5% improvement) while minimally impacting or even positively enhancing their helpfulness, outperforming various prompting methods. The dataset and code have both been open-sourced.

Mingyu Cao, Gen Li, Jie Ji et al.

Mixture-of-Experts (MoE) has garnered significant attention for its ability to scale up neural networks while utilizing the same or even fewer active parameters. However, MoE does not alleviate the massive memory requirements of networks, which limits their practicality in real-world applications, especially in the era of large language models (LLMs). While recent work explores the possibility of removing entire layers of MoE to reduce memory, the performance degradation is still notable. In this paper, we propose ConDense-MoE (CD-MoE), which, instead of dropping the entire MoE layer, condenses the large, sparse MoE layer into a smaller, denser layer with only a few experts activated for all tokens, while maintaining hardware friendliness. Our approach is specifically designed for fine-grained MoE with shared experts, where Feed-Forward Networks are split into many small experts, with certain experts isolated to serve as shared experts that are always activated, such as DeepSeekMoE and QwenMoE. We demonstrate the effectiveness of our method. Specifically, for the DeepSeekMoE-16B model, our approach maintains 90% of the average accuracy while reducing memory usage by 27.5% and increasing inference speed by 1.26 times. Moreover, we show that by applying lightweight expert fine-tuning -- only to the condensed layers -- and using 5 hours on a single 80G A100 GPU, we can successfully recover 98% of the original performance. Our code is available at: https://github.com/duterscmy/CD-MoE/tree/main.

Guinan Su, Li Shen, Lu Yin et al.

Large language models (LLMs) have shown remarkable capabilities in language understanding and generation. However, such impressive capability typically comes with a substantial model size, which presents significant challenges in deployment and inference. While structured pruning of model parameters offers a promising way to reduce computational costs at deployment time, current methods primarily focus on single model pruning. In this work, we develop a novel strategy to compress models by strategically combining or merging layers from finetuned model variants, which preserves the original model's abilities by aggregating capabilities accentuated in different finetunes. We pose the optimal tailoring of these LLMs as a zero-order optimization problem, adopting a search space that supports three different operations: (1) Layer removal, (2) Layer selection from different candidate models, and (3) Layer merging. Our experiments demonstrate that this approach leads to competitive model pruning, for example, for the Llama2-13B model families, our compressed models maintain approximately 97.3\% of the original performance while removing $\sim25\%$ of parameters, significantly outperforming previous state-of-the-art methods. The code is available at https://github.com/Guinan-Su/auto-merge-llm.

Zihan Wang, Rui Pan, Jiarui Yao et al.

We propose Chain-of-Experts (CoE), a new Mixture-of-Experts (MoE) architecture that introduces sequential expert communication within each layer. Unlike traditional MoE models, where experts operate independently in parallel, CoE processes tokens iteratively across a chain of experts inside a layer. To support dynamic expert selection across iterations, CoE employs a dedicated router at each iteration step within a layer. This design allows tokens to re-evaluate and select different experts during each iteration, rather than being statically assigned. As a result, CoE introduces a flexible routing mechanism that increases the diversity of expert combinations and enriches the model's representational capacity. CoE demonstrates improved performance under fixed compute: on math reasoning tasks, it reduces validation loss from 1.20 to 1.12 compared to a standard MoE. Beyond performance, CoE offers a new scaling axis: depth through expert iteration, which complements conventional width/depth scaling. For example, using 2x iterations matches the performance of 3x expert selections (in width), while reducing memory usage by 17.6-42% relative to other scaling strategies. Our analysis reveals that CoE's benefits stem from its iterative residual structure and enhanced expert specialization empowered by iterative routing, which together unlock more expressive representations. Code is available at https://github.com/ZihanWang314/coe.

Qiao Xiao, Alan Ansell, Boqian Wu et al.

Large language models (LLMs) have achieved remarkable success across various tasks but face deployment challenges due to their massive computational demands. While post-training pruning methods like SparseGPT and Wanda can effectively reduce the model size, but struggle to maintain model performance at high sparsity levels, limiting their utility for downstream tasks. Existing fine-tuning methods, such as full fine-tuning and LoRA, fail to preserve sparsity as they require updating the whole dense metrics, not well-suited for sparse LLMs. In this paper, we propose Sparsity Evolution Fine-Tuning (SEFT), a novel method designed specifically for sparse LLMs. SEFT dynamically evolves the sparse topology of pruned models during fine-tuning, while preserving the overall sparsity throughout the process. The strengths of SEFT lie in its ability to perform task-specific adaptation through a weight drop-and-grow strategy, enabling the pruned model to self-adapt its sparse connectivity pattern based on the target dataset. Furthermore, a sensitivity-driven pruning criterion is employed to ensure that the desired sparsity level is consistently maintained throughout fine-tuning. Our experiments on various LLMs, including LLaMA families, DeepSeek, and Mistral, across a diverse set of benchmarks demonstrate that SEFT achieves stronger performance while offering superior memory and time efficiency compared to existing baselines. Our code is publicly available at: https://github.com/QiaoXiao7282/SEFT.

Xinglei Wang, Tao Cheng, Stephen Law et al.

Existing methods for learning urban space representations from Point-of-Interest (POI) data face several limitations, including issues with geographical delineation, inadequate spatial information modelling, underutilisation of POI semantic attributes, and computational inefficiencies. To address these issues, we propose CaLLiPer (Contrastive Language-Location Pre-training), a novel representation learning model that directly embeds continuous urban spaces into vector representations that can capture the spatial and semantic distribution of urban environment. This model leverages a multimodal contrastive learning objective, aligning location embeddings with textual POI descriptions, thereby bypassing the need for complex training corpus construction and negative sampling. We validate CaLLiPer's effectiveness by applying it to learning urban space representations in London, UK, where it demonstrates 5-15% improvement in predictive performance for land use classification and socioeconomic mapping tasks compared to state-of-the-art methods. Visualisations of the learned representations further illustrate our model's advantages in capturing spatial variations in urban semantics with high accuracy and fine resolution. Additionally, CaLLiPer achieves reduced training time, showcasing its efficiency and scalability. This work provides a promising pathway for scalable, semantically rich urban space representation learning that can support the development of geospatial foundation models. The implementation code is available at https://github.com/xlwang233/CaLLiPer.

Tianhao Chen, Xin Xu, Lu Yin et al.

Transformer architectures serve as the backbone for most modern Large Language Models, therefore their pretraining stability and convergence speed are of central concern. Motivated by the logical dependency of sequentially stacked layers, we propose Progressive Residual Warmup (ProRes) for language model pretraining. ProRes implements an "early layer learns first" philosophy by multiplying each layer's residual with a scalar that gradually warms up from 0 to 1, with deeper layers taking longer warmup steps. In this way, deeper layers wait for early layers to settle into a more stable regime before contributing to learning. We demonstrate the effectiveness of ProRes through pretraining experiments across various model scales, as well as normalization and initialization schemes. Comprehensive analysis shows that ProRes not only stabilizes pretraining but also introduces a unique optimization trajectory, leading to faster convergence, stronger generalization and better downstream performance. Our code is available at https://github.com/dandingsky/ProRes.

Jiaxi Li, Lu Yin, Li Shen et al.

While large language models (LLMs) have achieved remarkable performance across a wide range of tasks, their massive scale incurs prohibitive computational and memory costs for pre-training from scratch. Recent studies have investigated the use of low-rank parameterization as a means of reducing model size and training cost. In this context, sparsity is often employed as a complementary technique to recover important information lost in low-rank compression by capturing salient features in the residual space. However, existing approaches typically combine low-rank and sparse components in a simplistic or ad hoc manner, often resulting in undesirable performance degradation compared to full-rank training. In this paper, we propose \textbf{LO}w-rank and \textbf{S}parse pre-\textbf{T}raining (\textbf{LOST}) for LLMs, a novel method that ingeniously integrates low-rank and sparse structures to enable effective training of LLMs from scratch under strict efficiency constraints. LOST applies singular value decomposition to weight matrices, preserving the dominant low-rank components, while allocating the remaining singular values to construct channel-wise sparse components to complement the expressiveness of low-rank training. We evaluate LOST on LLM pretraining ranging from 60M to 7B parameters. Our experiments show that LOST achieves competitive or superior performance compared to full-rank models, while significantly reducing both memory and compute overhead. Moreover, Code is available at \href{https://github.com/JiaxiLi1/LOST-Low-rank-and-Sparse-Training-for-Large-Language-Models}{LOST Repo}

Tianhao Chen, Xin Xu, Zijing Liu et al.

Modern Large Language Models, such as the LLaMA, Qwen and DeepSeek series, predominantly adopt the Pre-LayerNorm (Pre-LN) Transformer architecture. While being stable during pretraining and scalable to large model sizes, Pre-LN suffers from an exponential growth in activation variance across layers, causing the shortcut to dominate over sub-layer outputs in the residual connection and limiting the learning capacity of deeper layers. To mitigate this issue, we propose Gradient-Preserving Activation Scaling (GPAS), a simple technique that can be used in combination with existing approaches. GPAS works by scaling down the intermediate activations while keeping their gradients unchanged. This leaves information in the activations intact, and avoids the gradient vanishing problem associated with gradient downscaling. Extensive experiments across various model sizes from 71M to 1B show that GPAS achieves consistent performance gains. Beyond enhancing Pre-LN Transformers, GPAS also shows promise in improving alternative architectures such as Sandwich-LN and DeepNorm, demonstrating its versatility and potential for improving training dynamics in a wide range of settings. Our code is available at https://github.com/dandingsky/GPAS.

Xin Xu, Tianhao Chen, Fan Zhang et al.

While slow-thinking large language models (LLMs) exhibit reflection-like reasoning, commonly referred to as the "aha moment:, their ability to generate informative critiques and refine prior solutions remains limited. In this paper, we introduce Double-Checker, a principled framework designed to enhance the reasoning capabilities of slow-thinking LLMs by fostering explicit self-critique and iterative refinement of their previous solutions. By fine-tuning on our curated 1,730 self-critical instances, Double-Checker empowers long-CoT LLMs to iteratively critique and refine their outputs during inference until they evaluate their solutions as correct under self-generated critiques. We validate the efficacy of Double-Checker across a comprehensive suite of reasoning benchmarks, demonstrating that iterative self-critique significantly enhances the reasoning capabilities of long-CoT LLMs. Notably, our Double-Checker increases the pass@1 performance on challenging AIME benchmarks from 4.4% to 18.2% compared to the original long-CoT LLMs. These results highlight a promising direction for developing more trustworthy and effective LLMs capable of structured self-critique. Our codes and data are available at https://github.com/XinXU-USTC/DoubleChecker

Vishal Thengane, Zhaochong An, Tianjin Huang et al.

Incremental Few-Shot (IFS) segmentation aims to learn new categories over time from only a few annotations. Although widely studied in 2D, it remains underexplored for 3D point clouds. Existing methods suffer from catastrophic forgetting or fail to learn discriminative prototypes under sparse supervision, and often overlook a key cue: novel categories frequently appear as unlabelled background in base-training scenes. We introduce SCOPE (Scene-COntextualised Prototype Enrichment), a plug-and-play background-guided prototype enrichment framework that integrates with any prototype-based 3D segmentation method. After base training, a class-agnostic segmentation model extracts high-confidence pseudo-instances from background regions to build a prototype pool. When novel classes arrive with few labelled samples, relevant background prototypes are retrieved and fused with few-shot prototypes to form enriched representations without retraining the backbone or adding parameters. Experiments on ScanNet and S3DIS show that SCOPE achieves SOTA performance, improving novel-class IoU by up to 6.98% and 3.61%, and mean IoU by 2.25% and 1.70%, respectively, while maintaining low forgetting. Code is available https://github.com/Surrey-UP-Lab/SCOPE.

Keyu Wang, Tian Lyu, Guinan Su et al.

Layer pruning has emerged as a widely adopted technique for improving the efficiency of large language models (LLMs). Although existing methods demonstrate strong performance retention on general knowledge tasks, their effect on long-chain reasoning, a more brittle yet crucial capability, remains largely unexplored. In this work, we study the impact of layer pruning on long-chain reasoning through the lens of test-time scaling, a key mechanism in modern LLMs that enables strong reasoning capacity by allocating more computation at inference time. With extensive experiments, we demonstrate that pruning even one or two layers can severely impair test-time scaling, with performance collapsing drastically on long reasoning benchmarks even when performance on knowledge-intensive and shallow reasoning tasks remains stable. Furthermore, we find that standard supervised fine-tuning remedies fail to recover test-time scaling once it has deteriorated. Through in-depth analyses, we identify the mechanisms underlying this fragility of test-time scaling and highlight the fundamental risks of applying layer pruning to reasoning-intensive LLMs. These findings call for a rethinking of layer pruning strategies and provide insights for developing methods that preserve the robustness of reasoning. We open-source the codebase in \href{https://github.com/keyu-wang-2002/Layer-Pruning-Harms-Inference-Scaling}{https://github.com/keyu-wang-2002/Layer-Pruning-Harms-Inference-Scaling}.

Lu Yin, Ziying Shi, Yinghao Wu et al.

Human motion capture with sparse inertial sensors has gained significant attention recently. However, existing methods almost exclusively rely on a template adult body shape to model the training data, which poses challenges when generalizing to individuals with largely different body shapes (such as a child). This is primarily due to the variation in IMU-measured acceleration caused by changes in body shape. To fill this gap, we propose Shape-aware Inertial Poser (SAIP), the first solution considering body shape differences in sparse inertial-based motion capture. Specifically, we decompose the sensor measurements related to shape and pose in order to effectively model their joint correlations. Firstly, we train a regression model to transfer the IMU-measured accelerations of a real body to match the template adult body model, compensating for the shape-related sensor measurements. Then, we can easily follow the state-of-the-art methods to estimate the full body motions of the template-shaped body. Finally, we utilize a second regression model to map the joint velocities back to the real body, combined with a shape-aware physical optimization strategy to calculate global motions on the subject. Furthermore, our method relies on body shape awareness, introducing the first inertial shape estimation scheme. This is accomplished by modeling the shape-conditioned IMU-pose correlation using an MLP-based network. To validate the effectiveness of SAIP, we also present the first IMU motion capture dataset containing individuals of different body sizes. This dataset features 10 children and 10 adults, with heights ranging from 110 cm to 190 cm, and a total of 400 minutes of paired IMU-Motion samples. Extensive experimental results demonstrate that SAIP can effectively handle motion capture tasks for diverse body shapes. The code and dataset are available at https://github.com/yinlu5942/SAIP.

Zihao Wang, Enneng Yang, Lu Yin et al.

Model merging leverages multiple finetuned expert models to construct a multi-task model with low cost, and is gaining increasing attention. However, as a growing number of finetuned models become publicly available, concerns about the safety of model merging have emerged. Unauthorized merging may infringe on developers' rights and risk leaking sensitive personal information. Most existing methods focus on detecting whether a merged model originates from a specific source model, but fail to effectively prevent illegal merging. In this paper, we propose MergeLock, an active protection mechanism that disrupts model parameters to render them unmergeable, thereby directly preventing unauthorized model merging. Specifically, leveraging the inherent symmetry of the attention mechanism in Transformer-based models, we randomly sample two pairs of invertible matrices and apply them to the Query-Key (QK) and Value-Output (VO) branches. This transformation keeps the model's output unchanged while pushing it away from the shared parameter space of other finetuned models. Extensive experiments across both vision and language tasks demonstrate that MergeLock can degrade the performance of merged models by over 95% when a protected model is involved in most cases, demonstrating its effectiveness. Moreover, we further demonstrate that merged models protected by MergeLock cannot be effectively recovered using low-cost restoration methods, further enhancing robustness against unauthorized merging. The code is available at https://github.com/hetailang/Merge-Lock.

Di He, Songjun Tu, Ajay Jaiswal et al.

Weight decay is a standard regularization technique for training large language models (LLMs). While it is common to assign a uniform decay rate to every layer, this approach overlooks the structural diversity of LLMs and the varying spectral properties across modules. In this paper, we introduce AlphaDecay, a simple yet effective method that adaptively assigns different weight decay strengths to each module of an LLM. Our approach is guided by Heavy-Tailed Self-Regularization (HT-SR) theory, which analyzes the empirical spectral density (ESD) of weight correlation matrices to quantify "heavy-tailedness." Modules exhibiting more pronounced heavy-tailed ESDs, reflecting stronger feature learning, are assigned weaker decay, while modules with lighter-tailed spectra receive stronger decay. Our method leverages tailored weight decay assignments to balance the module-wise differences in spectral properties, leading to improved performance. Extensive pre-training tasks with various model sizes from 60M to 1B demonstrate that AlphaDecay achieves better perplexity and generalization than conventional uniform decay and other adaptive decay baselines. Our code is available at https://github.com/hed-ucas/AlphaDecay.

Xinglei Wang, Tao Cheng, Stephen Law et al.

Predicting individuals' next locations is a core task in human mobility modelling, with wide-ranging implications for urban planning, transportation, public policy and personalised mobility services. Traditional approaches largely depend on location embeddings learned from historical mobility patterns, limiting their ability to encode explicit spatial information, integrate rich urban semantic context, and accommodate previously unseen locations. To address these challenges, we explore the application of CaLLiPer -- a multimodal representation learning framework that fuses spatial coordinates and semantic features of points of interest through contrastive learning -- for location embedding in individual mobility prediction. CaLLiPer's embeddings are spatially explicit, semantically enriched, and inductive by design, enabling robust prediction performance even in scenarios involving emerging locations. Through extensive experiments on four public mobility datasets under both conventional and inductive settings, we demonstrate that CaLLiPer consistently outperforms strong baselines, particularly excelling in inductive scenarios. Our findings highlight the potential of multimodal, inductive location embeddings to advance the capabilities of human mobility prediction systems. We also release the code and data (https://github.com/xlwang233/Into-the-Unknown) to foster reproducibility and future research.

Lu Yin, Gen Li, Meng Fang et al.

Sparse training has received an upsurging interest in machine learning due to its tantalizing saving potential for the entire training process as well as inference. Dynamic sparse training (DST), as a leading sparse training approach, can train deep neural networks at high sparsity from scratch to match the performance of their dense counterparts. However, most if not all DST prior arts demonstrate their effectiveness on unstructured sparsity with highly irregular sparse patterns, which receives limited support in common hardware. This limitation hinders the usage of DST in practice. In this paper, we propose Channel-aware dynamic sparse (Chase), which for the first time seamlessly translates the promise of unstructured dynamic sparsity to GPU-friendly channel-level sparsity (not fine-grained N:M or group sparsity) during one end-to-end training process, without any ad-hoc operations. The resulting small sparse networks can be directly accelerated by commodity hardware, without using any particularly sparsity-aware hardware accelerators. This appealing outcome is partially motivated by a hidden phenomenon of dynamic sparsity: off-the-shelf unstructured DST implicitly involves biased parameter reallocation across channels, with a large fraction of channels (up to 60%) being sparser than others. By progressively identifying and removing these channels during training, our approach translates unstructured sparsity to channel-wise sparsity. Our experimental results demonstrate that Chase achieves 1.7 X inference throughput speedup on common GPU devices without compromising accuracy with ResNet-50 on ImageNet. We release our codes in https://github.com/luuyin/chase.

Tianjin Huang, Lu Yin, Zhenyu Zhang et al.

This paper reveals a new appeal of the recently emerged large-kernel Convolutional Neural Networks (ConvNets): as the teacher in Knowledge Distillation (KD) for small-kernel ConvNets. While Transformers have led state-of-the-art (SOTA) performance in various fields with ever-larger models and labeled data, small-kernel ConvNets are considered more suitable for resource-limited applications due to the efficient convolution operation and compact weight sharing. KD is widely used to boost the performance of small-kernel ConvNets. However, previous research shows that it is not quite effective to distill knowledge (e.g., global information) from Transformers to small-kernel ConvNets, presumably due to their disparate architectures. We hereby carry out a first-of-its-kind study unveiling that modern large-kernel ConvNets, a compelling competitor to Vision Transformers, are remarkably more effective teachers for small-kernel ConvNets, due to more similar architectures. Our findings are backed up by extensive experiments on both logit-level and feature-level KD ``out of the box", with no dedicated architectural nor training recipe modifications. Notably, we obtain the \textbf{best-ever pure ConvNet} under 30M parameters with \textbf{83.1\%} top-1 accuracy on ImageNet, outperforming current SOTA methods including ConvNeXt V2 and Swin V2. We also find that beneficial characteristics of large-kernel ConvNets, e.g., larger effective receptive fields, can be seamlessly transferred to students through this large-to-small kernel distillation. Code is available at: \url{https://github.com/VITA-Group/SLaK}.

Shiwei Liu, Tianlong Chen, Xiaohan Chen et al.

Works on lottery ticket hypothesis (LTH) and single-shot network pruning (SNIP) have raised a lot of attention currently on post-training pruning (iterative magnitude pruning), and before-training pruning (pruning at initialization). The former method suffers from an extremely large computation cost and the latter usually struggles with insufficient performance. In comparison, during-training pruning, a class of pruning methods that simultaneously enjoys the training/inference efficiency and the comparable performance, temporarily, has been less explored. To better understand during-training pruning, we quantitatively study the effect of pruning throughout training from the perspective of pruning plasticity (the ability of the pruned networks to recover the original performance). Pruning plasticity can help explain several other empirical observations about neural network pruning in literature. We further find that pruning plasticity can be substantially improved by injecting a brain-inspired mechanism called neuroregeneration, i.e., to regenerate the same number of connections as pruned. We design a novel gradual magnitude pruning (GMP) method, named gradual pruning with zero-cost neuroregeneration (\textbf{GraNet}), that advances state of the art. Perhaps most impressively, its sparse-to-sparse version for the first time boosts the sparse-to-sparse training performance over various dense-to-sparse methods with ResNet-50 on ImageNet without extending the training time. We release all codes in https://github.com/Shiweiliuiiiiiii/GraNet.

Shiwei Liu, Lu Yin, Decebal Constantin Mocanu et al.

In this paper, we introduce a new perspective on training deep neural networks capable of state-of-the-art performance without the need for the expensive over-parameterization by proposing the concept of In-Time Over-Parameterization (ITOP) in sparse training. By starting from a random sparse network and continuously exploring sparse connectivities during training, we can perform an Over-Parameterization in the space-time manifold, closing the gap in the expressibility between sparse training and dense training. We further use ITOP to understand the underlying mechanism of Dynamic Sparse Training (DST) and indicate that the benefits of DST come from its ability to consider across time all possible parameters when searching for the optimal sparse connectivity. As long as there are sufficient parameters that have been reliably explored during training, DST can outperform the dense neural network by a large margin. We present a series of experiments to support our conjecture and achieve the state-of-the-art sparse training performance with ResNet-50 on ImageNet. More impressively, our method achieves dominant performance over the overparameterization-based sparse methods at extreme sparsity levels. When trained on CIFAR-100, our method can match the performance of the dense model even at an extreme sparsity (98%). Code can be found https://github.com/Shiweiliuiiiiiii/In-Time-Over-Parameterization.

Shaojie Zhuang, Lu Yin, Guangshun Wei et al.

Automatic tooth segmentation and identification from intra-oral scanned 3D models are fundamental problems in digital dentistry, yet most existing approaches rely on task-specific 3D neural networks trained with densely annotated datasets, resulting in high annotation cost and limited generalization to scans from unseen sources. Thus, we propose TSegAgent, which addresses these challenges by reformulating dental analysis as a zero-shot geometric reasoning problem rather than a purely data-driven recognition task. The key idea is to combine the representational capacity of general-purpose foundation models with explicit geometric inductive biases derived from dental anatomy. Instead of learning dental-specific features, the proposed framework leverages multi-view visual abstraction and geometry-grounded reasoning to infer tooth instances and identities without task-specific training. By explicitly encoding structural constraints such as dental arch organization and volumetric relationships, the method reduces uncertainty in ambiguous cases and mitigates overfitting to particular shape distributions. Experimental results demonstrate that this reasoning-oriented formulation enables accurate and reliable tooth segmentation and identification with low computational and annotation cost, while exhibiting strong generalization across diverse and previously unseen dental scans.

Ajay Jaiswal, Bodun Hu, Lu Yin et al.

Autoregressive Large Language Models (e.g., LLaMa, GPTs) are omnipresent achieving remarkable success in language understanding and generation. However, such impressive capability typically comes with a substantial model size, which presents significant challenges for autoregressive token-by-token generation. To mitigate computation overload incurred during generation, several early-exit and layer-dropping strategies have been proposed. Despite some promising success due to the redundancy across LLMs layers on metrics like Rough-L/BLUE, our careful knowledge-intensive evaluation unveils issues such as generation collapse, hallucination of wrong facts, and noticeable performance drop even at the trivial exit ratio of 10-15% of layers. We attribute these errors primarily to ineffective handling of the KV cache through state copying during early-exit. In this work, we observed the saturation of computationally expensive feed-forward blocks of LLM layers and proposed FFN-SkipLLM, which is a novel fine-grained skip strategy of autoregressive LLMs. More specifically, FFN-SkipLLM is an input-adaptive feed-forward skipping strategy that can skip 25-30% of FFN blocks of LLMs with marginal change in performance on knowledge-intensive generation tasks without any requirement to handle KV cache. Our extensive experiments and ablation across benchmarks like MT-Bench, Factoid-QA, and variable-length text summarization illustrate how our simple and ease-at-use method can facilitate faster autoregressive decoding.

Mingyu Cao, Alvaro Correia, Christos Louizos et al.

Diffusion Language Models (DLMs) generate text by iteratively denoising a masked sequence, repeatedly deciding which positions to commit at each step. Standard decoding follows a greedy rule: unmask the most confident positions, yet this local choice can lock the model into a suboptimal unmasking order, especially on reasoning-heavy prompts. We present SOAR, a training-free decoding algorithm that adapts its behavior to the model's uncertainty. When confidence is low, SOAR briefly widens the search over alternative unmasking decisions to avoid premature commitments; when confidence is high, it collapses the search and decodes many positions in parallel to reduce the number of denoising iterations. Across mathematical reasoning and code generation benchmarks (GSM8K, MBPP, HumanEval) on Dream-7B and LLaDA-8B, SOAR improves generation quality while maintaining competitive inference speed, offering a practical way to balance quality and efficiency in DLM decoding.

Ziyang Chen, Zhenxuan Huang, Yile Wang et al.

Traditional sentence embedding methods employ token-level contrastive learning on non-generative pre-trained models. Recently, there have emerged embedding methods based on generative large language models (LLMs). These methods either rely on fixed prompt templates or involve modifications to the model architecture. The former lacks further optimization of the model and results in limited performance, while the latter alters the internal computational mechanisms of the model, thereby compromising its generative capabilities. We propose SemPA, a novel approach that boosts the sentence representations while preserving the generative ability of LLMs via semantic preference alignment. We leverage sentence-level Direct Preference Optimization (DPO) to efficiently optimize LLMs on a paraphrase generation task, where the model learns to discriminate semantically equivalent sentences while preserving inherent generative capacity. Theoretically, we establish a formal connection between DPO and contrastive learning under the Plackett-Luce model framework. Empirically, experimental results on both semantic textual similarity tasks and various benchmarks for LLMs show that SemPA achieves better semantic representations without sacrificing the inherent generation capability of LLMs.