Large Language Models (LLMs) with billions of parameters have transformed AI applications but require immense computational and memory resources during inference. Adaptive sparse activation inference, which activates only a small number of neurons for each token, offers a novel way to accelerate model inference without degrading performance, showing great potential for resource-constrained hardware devices. Yet existing token-level MLP-based methods frequently alter activation maps, reducing efficiency gains. In this paper, we introduce \textbf{SparseInfer}, an MLP-free adaptive sparse activation inference method based on sentence-level prediction. We first propose the concept of core neurons and empirically demonstrate that, for an input sentence, LLMs only need the core neurons to maintain performance. Remarkably, we discovered that core neurons exhibit both stability and similarity in relation to the sentence's semantics—an insight overlooked by previous studies. Building on this finding, we design two semantic-based methods for predicting core neurons to fit different input scenarios, which enables core neurons to be determined during the pre-filling stage and fixed during the encoding stage. Our experiments verify SparseInfer exhibits good performance in various tasks and achieve 10.33$\times$ speed up.
论文检索
输入标题、作者或关键词，从 100,903 篇学术成果中精准定位
Optimization · Non-Convex
Sign-based optimization methods, such as SignSGD and Signum, have become essential for modern Deep Learning due to their 1) high performance 2) low memory footprint and 3) communication efficiency. Despite their success, these methods suffer from distinct limitations in the terminal phase of training: they decouple update mechanisms from gradient magnitudes and fail to account for parameter heterogeneity, often leading to oscillation rather than convergence. While switching to SGD represents a potential remedy, a naive "hard" switch is poorly useful due to learning rate mismatches, momentum buffer suboptimality, and the assumption of uniform parameter dynamics. In this work, we propose SoftSignum, a novel optimization method that implements a principled, smooth transition mechanism from sign-based updates to SGD, which adapts to individual parameter sensitivities. We provide a generalized theoretical framework guaranteeing convergence in stochastic non-convex settings relevant to Deep Learning and demonstrate empirically that SoftSignum effectively handles parameter heterogeneity, yielding superior convergence across diverse tasks, including LLM pretraining, compared to standard sign-based baselines.
Applications · Neuroscience, Cognitive Science
Credit assignment, the process of determining how changes in individual neurons and synapses influence a network’s output, is central to learning in brains and machines. Noise correlation-based methods, which estimate gradients by correlating perturbations of activity with changes in output, provide a biologically plausible solution to credit assignment but scales poorly as accurately estimating the Jacobian requires that the number of perturbations scale with network size. Moreover, isotropic noise conflicts with neurobiological observations that neural activity lies on a low-dimensional manifold. To address these drawbacks, we propose *neural manifold noise correlation* (NMNC), which performs credit assignment using perturbations restricted to the neural manifold. We show theoretically and empirically that the Jacobian row space aligns with the neural manifold in trained networks, and that manifold dimensionality scales slowly with network size. NMNC substantially improves performance and sample efficiency over vanilla noise correlation in convolutional networks trained on CIFAR-10, ImageNet-scale models, and recurrent networks. NMNC also yields representations more similar to the primate visual system than vanilla noise correlation. These findings offer a mechanistic hypothesis for how biological circuits could support credit assignment, and suggest that biologically inspired constraints may enable, rather than limit, effective learning at scale.
Large language models (LLMs) perform inference by following a fixed depth and order, non-recurrent execution of all layers. We reveal the wide existence of training-free, flexible, dynamic "program-of-layers (PoLar)", where pretrained layers can be packed as modules and then skipped or looped to form a customized program for each input. For most inputs, substantially shorter program executions can achieve the same or better accuracy, while incorrect predictions of the original LLM can be corrected by alternative programs with fewer layers. These observations indicate that inference admits multiple valid latent computations beyond the standard forward pass. To efficiently achieve PoLar in practice, we propose a lightweight PoLar prediction network, which learns to generate execution programs that dynamically skip or repeat pretrained layers for each input. Experiments on mathematical reasoning benchmarks demonstrate that PoLar consistently improves accuracy over standard inference and prior dynamic-depth methods, often while executing fewer layers, and that these gains persist under out-of-distribution evaluation. Our results suggest that fixed-depth execution captures only a narrow subset of an LLM’s latent reasoning capacity.
Deep Learning · Graph Neural Networks
Higher-order structures are powerful relational modeling tools, yet existing spectral operators decompose topology into separate ranks, leaving practitioners to fuse information back to vertices through ad-hoc choices. We introduce _Collapsed Effective Operators_, which marginalize higher-order structures into a single vertex-level operator via Schur complementation of a graded Laplacian. This yields a dense operator that encodes long-range interactions mediated by topology and is applicable to arbitrary higher-order constructs. We show it preserves positive semi-definiteness with a strict spectral upper bound relative to the rank-0 Laplacian, effectively lowering system energy under higher-order connectivity. Empirically, our operator significantly improves spectral clustering, enables diffusion over topological structures, and accelerates the processing of higher-order structures with neural networks.
Applications · Time Series
Standard attention mechanisms in transformers employ static token representations that remain unchanged across all pair-wise computations in each layer. This limits their representational alignment with the potentially diverse dynamics of each token-pair interaction. While they excel in domains with relatively homogeneous relationships, standard attention may be inadequate in capturing heterogeneous inter-channel dependencies of multivariate time series (MTS) data where different channel-pair interactions within a single system may be governed by entirely different physical laws or temporal dynamics. To better align the attention mechanism for such domain phenomena, we propose attention with dynamic relational priming (prime attention). Prime attention modulates token representations for each token-pair, optimizing each pair-wise interaction for that specific relationship. Our results demonstrate that prime attention consistently outperforms standard attention across benchmarks, achieving up to 6.5\% improvement in forecasting accuracy. In addition, prime attention achieves comparable performance using up to 40\% less sequence length compared to standard attention, demonstrating its superior relational modeling capabilities and potential for data efficiency.
Deep Learning · Large Language Models
Mixture-of-Experts(MoE) Vision-Language Models(VLMs) offer remarkable performance but incur prohibitive memory and computational costs, making compression essential. Post-Training Quantization (PTQ) is an effective training-free technique to address the massive memory and computation overhead. Existing quantization paradigms fall short as they are oblivious to two critical forms of heterogeneity: the inherent discrepancy between vision and language tokens, and the non-uniform contribution of different experts. To bridge this gap, we introduce Visual Expert Quantization (VEQ), a dual-aware quantization framework designed to simultaneously accommodate cross-modal differences and heterogeneity between experts. Specifically, VEQ incorporates 1)**Modality-expert-aware Quantization**, which utilizes expert activation frequency to prioritize error minimization for pivotal experts, and 2)**Modality-affinity-aware Quantization**, which constructs an enhanced Hessian matrix by integrating token-expert affinity with modality information to guide the calibration process. Extensive experiments across diverse benchmarks verify that VEQ consistently outperforms state-of-the-art baselines. Specifically, under the W3A16 configuration, our method achieves significant average accuracy gains of 2.04\% on Kimi-VL and 3.09\% on Qwen3-VL compared to the previous SOTA quantization methods, demonstrating superior robustness across various multi-modal tasks.
General Machine Learning · Evaluation
Large Language Models (LLMs) excel at long-context understanding but exhibit significant limitations in long-form generation. Existing studies primarily focus on single-generation quality, generally overlooking the volatility of the output (i.e., the inconsistency in length and content across multiple generations). This volatility not only leads to significant computational costs but also severely impacts the models' reliable application. To address this gap, our work unfolds in three stages: benchmarking, probing, and mitigation. We first propose the VOlatility in Long-form Text Benchmark (VOLTBench), a novel heterogeneous-task benchmark designed to systematically quantify the length volatility of long-form generation. Subsequently, by analyzing attention traces, we conduct an in-depth probe to identify several common internal patterns that cause this volatility. Finally, to mitigate long-form output volatility, we propose SELB (Structural Enforcement via Logits Boosting), a lightweight decoding-stage optimization strategy, designed to significantly enhance both the length accuracy and stability of long-form generation without additional training. Extensive experiments on VOLTBench provide the first systematic confirmation of severe long-form output instability in mainstream models and validate that our proposed method successfully improves the mean output length of the base model by 148% and reduces the length volatility by 69%, while maintaining high generation quality.
Deep Learning · Generative Models and Autoencoders
One emerging approach to mitigating data heterogeneity in Federated Learning (FL) is to employ diffusion models to generate synthetic data for clients, thereby aligning local data distributions with the global distribution. Prior work has primarily focused on balance-oriented augmentation, which assumes a balanced global class distribution and thus generates samples of rare classes to rebalance each client's local dataset. However, in practice, global data distributions are often inherently imbalanced. Moreover, privacy constraints in FL hinder the server’s ability to accurately estimate the global distribution, rendering balance-oriented augmentation suboptimal. This raises a key, underexplored challenge: How can synthetic data be generated and selected to align local distributions with the true, yet unknown, global distribution? Our key insight is that a model’s performance implicitly reflects the data distribution it has been trained on. Based on this observation, we use the performance discrepancy between local and global models to identify the regions where each client’s local dataset is lacking, and generate corresponding samples for clients. Furthermore, we adapt the diffusion model via preference optimization, enabling it to generate data that better aligns with the true global distribution. Extensive experiments on multiple benchmarks demonstrate that FedPDG outperforms state-of-the-art methods, achieving up to 3.82\% improvement.
Reinforcement Learning · Multi-agent
Uncovering the causal mechanisms of educational social dynamics is critical for designing effective pedagogical policies. However, traditional methods face a fundamental dilemma. vational studies often lack causal power, while controlled experiments are ethically prohibitive. While LLM-powered multi-agent simulations offer a scalable in silico alternative, current approaches often fail to support rigorous experimentation due to shallow psychological grounding and unquantifiable interactions. To address this, we introduce EduMirror, a multi-agent simulator for the scientific study of educational social dynamics. EduMirror employs a value-driven cognitive architecture for agents that grounds agent behaviors in social value and intrinsic motivation, coupled with a dual-track measurement protocol that utilizes LLMs to quantify both overt actions and latent psychological states. We validate the realism and usability of our platform through case studies on school bullying and group cooperation. The results show that EduMirror generates realistic social phenomena aligned with established theories and measurable by empirical criteria. These properties enable structured in-silico educational research. Results demonstrate that EduMirror generates dynamics aligned with established theories, providing a robust tool for hypothesis testing in educational science.
We argue that many Anthropomorphized Misalignment Research (AMR) studies need stronger evidence to ensure that they can provide a robust foundation for critical safety decisions, such as model deployment and regulation. By evaluating failure modes across different misalignment concepts, such as deception, emergent misalignment, and sycophancy, we show how conceptual ambiguity, non-robust datasets and experimental design, and insufficient causal interventions can lead to overinterpretation of model behaviors. This position paper aims to offer guidance on evidentiary considerations that can help improve methodological rigor in AMR. To achieve this, we provide a clear call to action through a proposed framework of evidence levels and a diagnostic checklist. These shared standards will enable more productive scientific discourse and ensure that claims about AI risks rest on solid empirical foundations.
Deep Learning · Large Language Models
Evaluating open-ended outputs from large language models (LLMs) remains challenging due to the absence of ground truth. We introduce TRACE (Toulmin-based Reasoning Assessment through Constructive Elements), a metric that analyzes Chain-of-Thought (CoT) reasoning processes. TRACE integrates Toulmin's argumentation theory with Flavell's metacognitive framework to assess reasoning structure. Experiments on 26.3K QA samples across 7 reasoning models show strong correlation with benchmark accuracy (r = 0.74). Furthermore, TRACE is effective as a reinforcement learning reward signal, outperforming accuracy-only baselines. These results suggest that TRACE serves as a complementary metric for evaluating open-ended outputs.
Social Aspects · Alignment
We argue that many Anthropomorphized Misalignment Research (AMR) studies need stronger evidence to ensure that they can provide a robust foundation for critical safety decisions, such as model deployment and regulation. By evaluating failure modes across different misalignment concepts, such as deception, emergent misalignment, and sycophancy, we show how conceptual ambiguity, non-robust datasets and experimental design, and insufficient causal interventions can lead to overinterpretation of model behaviors. This position paper aims to offer guidance on evidentiary considerations that can help improve methodological rigor in AMR. To achieve this, we provide a clear call to action through a proposed framework of evidence levels and a diagnostic checklist. These shared standards will enable more productive scientific discourse and ensure that claims about AI risks rest on solid empirical foundations.
Applications · Health / Medicine
While Large Language Models (LLMs) have achieved remarkable success in dyadic (one-on-one) instruction, they face significant challenges in One-to-Many alignment, such as clinical ward rounds, where an instructor must simultaneously guide a diverse group of trainees. Current models often suffer from context dilution and goal misalignment, failing to balance individual scaffolding with collective learning progress. To address this, we introduce ClinEdu, a multi-agent pedagogical simulator that model the complexity of group dynamics. Leveraging this platform, we construct ClinTeach, a large-scale dataset of Socratic teaching dialogues, and propose ClinTutor-R1, the first multimodal agent explicitly architected to achieve one-to-many alignment in clinical education, employing an explicit internal thinking mechanism to model both individual belief states and group consensus. We validate our framework through a comprehensive protocol covering both standard static benchmarks and rigorous in-situ interactive evaluation within ClinEdu. Experimental results demonstrate that ClinTutor-R1 outperforms base models by over 20% and achieves parity with proprietary reasoning models , while exhibiting exceptional scalability in maintaining instructional quality across expanding student cohorts.
Deep Learning · Large Language Models
Improving the reasoning capabilities of large language models (LLMs) typically relies either on the model's ability to sample a correct solution to be reinforced or the existence of a stronger model able to solve the problem. However, many difficult problems remain intractable for even current frontier models, preventing the extraction of valid training signals. A promising alternative is to leverage high-quality expert human solutions, yet naive imitation of this data fails because it is fundamentally out-of-distribution: expert solutions are typically didactic, containing implicit reasoning gaps intended for human readers rather than computational models. Furthermore, high-quality expert solutions are expensive, necessitating generalizable sample-efficient training methods. We propose Distribution Aligned Imitation Learning (DAIL), a two-step method that bridges the distributional gap by first transforming expert solutions into detailed, in-distribution reasoning traces and then applying a contrastive objective to focus learning on expert insights and methodologies. We find that DAIL can leverage fewer than 1000 high-quality expert solutions to achieve 10–25\% pass@$k$ gains on Qwen2.5-Instruct and Qwen3 models, improve reasoning efficiency by $2\times$ to $4\times$, and enable out-of-domain generalization.
General Machine Learning · Data
Generative tabular augmentation is appealing in data-scarce domains, yet the prevailing focus on distributional fidelity does not reliably translate into better downstream models. We formalize a *fidelity-utility gap*: common generative objectives prioritize distributional plausibility, whereas augmentation succeeds only when injected samples reduce the current learner's held-out evaluation loss. This gap motivates learning not just how to generate, but what to generate and when to inject as training evolves. We propose TAP (Tabular Augmentation Policy), which couples diffusion inpainting with a lightweight, learner-conditioned policy to steer generation toward high-utility regions and controls safe injection via explicit gating and conservative windowed commitment. Under severe data scarcity, TAP consistently outperforms strong generative baselines on seven real-world datasets, improving classification accuracy by up to 15.6 percentage points and reducing regression RMSE by up to 32%.
Deep Learning · Generative Models and Autoencoders
We present Causal-Adapter, a modular framework that adapts frozen text-to-image diffusion backbones for counterfactual image generation. Our method enables causal interventions on target attributes while preserving all other aspects of the image, including the core identity. In contrast to prior approaches that rely on prompt engineering without explicit causal structure, Causal-Adapter leverages structural causal modeling augmented with two attribute regularization strategies: prompt-aligned injection, which aligns causal attributes with textual embeddings for precise semantic control, and a conditioned token contrastive loss to disentangle attribute factors and reduce spurious correlations. Causal-Adapter achieves state-of-the-art results on synthetic and real-world datasets, outperforming other baselines in effectiveness, composition, realism, and minimality. These results demonstrate that the approach enables efficient, robust, and generalizable counterfactual image editing with faithful attribute modification and strong preservation of core identity.
Deep Learning · Large Language Models
Large Reasoning Models (LRMs) solve complex tasks by generating long Chain-of-Thought (CoT) sequences; however, the emergent dynamics governing reasoning trajectories are not well understood and can lead to inconsistencies and reasoning pathologies. In this work, we propose to approximate LRM's emerging hierarchical reasoning dynamics as a trajectory within a Finite State Machine (FSM) transitioning among six abstract cognitive states. We demonstrate that these states and transitions can be captured in the latent state of the model. We believe that this representation can have different applications in the interpretability and optimization of LRM models. For example, by analyzing the topology of these transitions, we identify statistical shifts in reasoning strategies that help identify effective reasoning chains from those that fail. To illustrate these potential advantages, we propose $Q$-Value guided steering, a training-free inference-time control method that treats reasoning as a planning problem. We estimate the long-horizon utility of state transitions and apply sparse, orthogonal activation steering at sentence boundaries to align the CoT generation with optimal reasoning policies. Experiments across four benchmarks (AIME25, MATH-500, GSM8k, and GPQA Diamond) using three state-of-the-art open reasoning models demonstrate that $Q$-Value steering policy achieves significant performance gains with "surgical'' efficiency, often requiring $25\times$ fewer interventions than greedy and weighted baselines, which suggests that reasoning can be effectively controlled by guiding high-level cognitive dynamics rather than micro-managing token generation.
Deep Learning · Large Language Models
Direct alignment methods are increasingly used to align large language models (LLMs) with human preferences. However, many real-world alignment problems involve multiple conflicting objectives, where naive aggregation of preferences can lead to unstable training and poor trade-offs. In particular, weighted loss methods may fail to identify update directions that simultaneously improve all objectives, and existing multi-objective approaches often rely on explicit reward models, introducing additional complexity and distorting user-specified preferences. The contributions of this paper are two-fold. First, we propose a **R**eward-free **A**lignment framework for **C**onflicted **O**bjectives (RACO) that directly leverages pairwise preference data and resolves gradient conflicts via a novel clipped variant of conflict-averse gradient descent. We provide convergence guarantees to Pareto-critical points that respect user-specified objective weights, and further show that clipping can strictly improve convergence rate in the two-objective setting. Second, we improve our method using some heuristics and conduct experiments to demonstrate the compatibility of the proposed framework for LLM alignment. Both qualitative and quantitative evaluations on multi-objective summarization and safety alignment tasks across multiple LLM families (Qwen 3, Llama 3, Gemma 3) show that our method consistently achieves better Pareto trade-offs compared to existing multi-objective alignment baselines.
Deep Learning · Attention Mechanisms
We characterize the pre-softmax attention matrix $\mathbf{QK^\top}$ in transformers as an associative memory matrix encoding pairwise associations between input features. By decomposing this matrix into its symmetric and skew-symmetric parts, we interpret the symmetric component as governing the structure of the *energy landscape*, and the skew-symmetric component as driving *circulation* on that landscape. Leveraging the energy formulation induced by the symmetric component, we derive Hopfield-style stability measures that quantify the stability of retrieved features. Empirically, we observe meaningful correlations between Hopfield-style stability measures and the fidelity$-$diversity trade-offs in generation. Finally, we propose a controllable knob to modulate this trade-off by directly modifying the circulation of the underlying dynamics.