Searcharxiv⌕ Search

arXiv subjects

Guodong Du

Publications and source records attributed to Guodong Du.

At least 19 recordsLinked to original sources
arXiv2026

Post-Training Shifts Confidence: A Three-Stage Analysis of How SFT, RL, and OPD Shape CoT Calibration

Large language models have made strong reasoning gains through supervised fine-tuning, reinforcement learning, and on-policy distillation, yet these post-training methods are usually evaluated only by final-answer accuracy. We study how they reshape confidence during reasoning. We introduce a three-stage calibration framework that evaluates confidence before, during, and after chain-of-thought generation, corresponding to difficulty estimation, early termination, and answer aggregation. Through a controlled comparison on mathematical reasoning benchmarks, we find that OPD provides the most useful pre-reasoning confidence, SFT gives the strongest online signal for early stopping, and RL produces the most reliable trace-level signal for aggregation. We further show that confidence reliability is position-dependent: RL confidence becomes informative after a path-commitment phase, while OPD confidence is useful early but can become inversely calibrated later. Based on this observation, we propose PosConf, a position-aware confidence strategy that uses confidence only from reliable relative-position intervals. PosConf improves RL answer aggregation by 6.1 points over majority voting and consistently improves OPD early stopping under tight token budgets, with gains up to 4.3 points by avoiding its later inverse-calibration region, showing that \emph{confidence in reasoning models should be used both stage-wise and position-awarely}. Our code is available at https://github.com/EIT-NLP/Post-Training-Calibration.

cs.CL↗
arXiv2026

Skill Weaving: Efficient LLM Improvement via Modular Skillpacks

Large language models increasingly require specialization across diverse domains, yet existing approaches struggle to balance multi-domain capacities with strict memory and inference constraints. In this work, we introduce SkillWeave, a modular improvement framework that enables LLMs to specialize under fixed memory budgets. SkillWeave partitions full capabilities of a general-purpose model into skillpacks -- lightweight, domain-specific delta modules -- that reorganize and refine the model's internal knowledge. For efficient deployment, SkillWeave integrates SkillZip to compress skillpacks into compact and inference-ready format, enabling strong multi-domain performance with low-latency execution. On multi-task and agentic benchmarks, a 9B SkillWeave model outperforms several baselines and even surpasses a 32B monolithic LLM, while achieving up to 4x speedup.

cs.AI↗
arXiv2026

Dynamic Model Merging Made Slim

Model merging enables the reuse of fine-tuned models without joint training or access to original data. Dynamic merging further improves flexibility by selectively activating task-relevant parameters and efficiently composing experts across multiple tasks. However, existing dynamic methods either maintain a full shared model with tiny experts or allocate excessive capacity to experts, leading to suboptimal accuracy--efficiency trade-offs. To address this, we propose DiDi-Merging, a slim dynamic merging framework that leverages differentiable rank allocation to balance shared and expert parameters. By formulating parameter budgeting as differentiable rank optimization in low-rank modules and introducing a data-free refinement step to recover task fidelity, DiDi-Merging matches prior dynamic baselines at only 1.24x the parameters of a single fine-tuned model and surpasses them at 1.4x, substantially more compact than methods requiring > 2x storage. DiDi-Merging applies across vision, language, and multimodal tasks.

cs.LG↗
arXiv2026

Vocabulary Hijacking in LVLMs: Unveiling Critical Attention Heads by Excluding Inert Tokens to Mitigate Hallucination

Large Vision-Language Models (LVLMs) have achieved remarkable progress in multimodal tasks, yet their reliability is persistently undermined by hallucinations-generating text that contradicts visual input. Recent studies often attribute these errors to inadequate visual attention. In this work, we analyze the attention mechanisms via the logit lens, uncovering a distinct anomaly we term Vocabulary Hijacking. We discover that specific visual tokens, defined as Inert Tokens, disproportionately attract attention. Crucially, when their intermediate hidden states are projected into the vocabulary space, they consistently decode to a fixed set of unrelated words (termed Hijacking Anchors) across layers, revealing a rigid semantic collapse. Leveraging this semantic rigidity, we propose Hijacking Anchor-Based Identification (HABI), a robust strategy to accurately localize these Inert Tokens. To quantify the impact of this phenomenon, we introduce the Non-Hijacked Visual Attention Ratio (NHAR), a novel metric designed to identify attention heads that remain resilient to hijacking and are critical for factual accuracy. Building on these insights, we propose Hijacking-Aware Visual Attention Enhancement (HAVAE), a training-free intervention that selectively strengthens the focus of these identified heads on salient visual content. Extensive experiments across multiple benchmarks demonstrate that HAVAE significantly mitigates hallucinations with no additional computational overhead, while preserving the model's general capabilities. Our code is publicly available at https://github.com/lab-klc/HAVAE.

cs.MM↗
arXiv2026

D-QRELO: Training- and Data-Free Delta Compression for Large Language Models via Quantization and Residual Low-Rank Approximation

Supervised Fine-Tuning (SFT) accelerates taskspecific large language models (LLMs) development, but the resulting proliferation of finetuned models incurs substantial memory overhead. Delta compression addresses this by retaining a single pre-trained LLM with multiple compressed delta weights. However, existing methods fail on models fine-tuned with largescale datasets. We find that larger SFT data scale amplifies delta parameter magnitude, singular values, and entropy, exacerbating compression errors. To tackle this, we propose DQRELO (Delta Compression via Quantization and Residual Low-Rank), a novel training- and data-free delta compression method. It combines coarse-grained one-bit quantization to capture the dominant structure of the delta, followed by compensated residual low-rank approximation to recover fine-grained details from the smaller residual error. Experiments on various LLMs spanning dense and MoE architectures across multiple domains under this challenging setting demonstrate that DQRELO outperforms existing methods. Moreover, we establish key design principles for delta compression through extensive empirical analysis, demonstrating how task difficulty, architecture, and layer positioning create predictable patterns that can guide optimal compression strategies in production systems.

cs.LG↗
arXiv2025

NeurIPT: Foundation Model for Neural Interfaces

Electroencephalography (EEG) has wide-ranging applications, from clinical diagnosis to brain-computer interfaces (BCIs). With the increasing volume and variety of EEG data, there has been growing interest in establishing foundation models (FMs) to scale up and generalize neural decoding. Despite showing early potential, applying FMs to EEG remains challenging due to substantial inter-subject, inter-task, and inter-condition variability, as well as diverse electrode configurations across recording setups. To tackle these open challenges, we propose NeurIPT, a foundation model developed for diverse EEG-based Neural Interfaces with a Pre-trained Transformer by capturing both homogeneous and heterogeneous spatio-temporal characteristics inherent in EEG signals. Temporally, we introduce Amplitude-Aware Masked Pretraining (AAMP), masking based on signal amplitude rather than random intervals, to learn robust representations across varying signal intensities beyond local interpolation. Moreover, this temporal representation is enhanced by a Progressive Mixture-of-Experts (PMoE) architecture, where specialized expert subnetworks are progressively introduced at deeper layers, adapting effectively to the diverse temporal characteristics of EEG signals. Spatially, NeurIPT leverages the 3D physical coordinates of electrodes, enabling effective transfer of embedding across varying EEG settings, and develops Intra-Inter Lobe Pooling (IILP) during fine-tuning to efficiently exploit regional brain features. Empirical evaluations across eight downstream BCI datasets, via fine-tuning, demonstrated NeurIPT consistently achieved state-of-the-art performance, highlighting its broad applicability and robust generalization. Our work pushes forward the state of FMs in EEG and offers insights into scalable and generalizable neural information processing systems.

cs.LG↗
arXiv2025

Escaping Stability-Plasticity Dilemma in Online Continual Learning for Motion Forecasting via Synergetic Memory Rehearsal

Deep neural networks (DNN) have achieved remarkable success in motion forecasting. However, most DNN-based methods suffer from catastrophic forgetting and fail to maintain their performance in previously learned scenarios after adapting to new data. Recent continual learning (CL) studies aim to mitigate this phenomenon by enhancing memory stability of DNN, i.e., the ability to retain learned knowledge. Yet, excessive emphasis on the memory stability often impairs learning plasticity, i.e., the capacity of DNN to acquire new information effectively. To address such stability-plasticity dilemma, this study proposes a novel CL method, synergetic memory rehearsal (SyReM), for DNN-based motion forecasting. SyReM maintains a compact memory buffer to represent learned knowledge. To ensure memory stability, it employs an inequality constraint that limits increments in the average loss over the memory buffer. Synergistically, a selective memory rehearsal mechanism is designed to enhance learning plasticity by selecting samples from the memory buffer that are most similar to recently observed data. This selection is based on an online-measured cosine similarity of loss gradients, ensuring targeted memory rehearsal. Since replayed samples originate from learned scenarios, this memory rehearsal mechanism avoids compromising memory stability. We validate SyReM under an online CL paradigm where training samples from diverse scenarios arrive as a one-pass stream. Experiments on 11 naturalistic driving datasets from INTERACTION demonstrate that, compared to non-CL and CL baselines, SyReM significantly mitigates catastrophic forgetting in past scenarios while improving forecasting accuracy in new ones. The implementation is publicly available at https://github.com/BIT-Jack/SyReM.

cs.LG↗
arXiv2025

Complementary Learning System Empowers Online Continual Learning of Vehicle Motion Forecasting in Smart Cities

Artificial intelligence underpins most smart city services, yet deep neural network (DNN) that forecasts vehicle motion still struggle with catastrophic forgetting, the loss of earlier knowledge when models are updated. Conventional fixes enlarge the training set or replay past data, but these strategies incur high data collection costs, sample inefficiently and fail to balance long- and short-term experience, leaving them short of human-like continual learning. Here we introduce Dual-LS, a task-free, online continual learning paradigm for DNN-based motion forecasting that is inspired by the complementary learning system of the human brain. Dual-LS pairs two synergistic memory rehearsal replay mechanisms to accelerate experience retrieval while dynamically coordinating long-term and short-term knowledge representations. Tests on naturalistic data spanning three countries, over 772,000 vehicles and cumulative testing mileage of 11,187 km show that Dual-LS mitigates catastrophic forgetting by up to 74.31\% and reduces computational resource demand by up to 94.02\%, markedly boosting predictive stability in vehicle motion forecasting without inflating data requirements. Meanwhile, it endows DNN-based vehicle motion forecasting with computation efficient and human-like continual learning adaptability fit for smart cities.

cs.LG↗
arXiv2025

H2C: Hippocampal Circuit-inspired Continual Learning for Lifelong Trajectory Prediction in Autonomous Driving

Deep learning (DL) has shown state-of-the-art performance in trajectory prediction, which is critical to safe navigation in autonomous driving (AD). However, most DL-based methods suffer from catastrophic forgetting, where adapting to a new distribution may cause significant performance degradation in previously learned ones. Such inability to retain learned knowledge limits their applicability in the real world, where AD systems need to operate across varying scenarios with dynamic distributions. As revealed by neuroscience, the hippocampal circuit plays a crucial role in memory replay, effectively reconstructing learned knowledge based on limited resources. Inspired by this, we propose a hippocampal circuit-inspired continual learning method (H2C) for trajectory prediction across varying scenarios. H2C retains prior knowledge by selectively recalling a small subset of learned samples. First, two complementary strategies are developed to select the subset to represent learned knowledge. Specifically, one strategy maximizes inter-sample diversity to represent the distinctive knowledge, and the other estimates the overall knowledge by equiprobable sampling. Then, H2C updates via a memory replay loss function calculated by these selected samples to retain knowledge while learning new data. Experiments based on various scenarios from the INTERACTION dataset are designed to evaluate H2C. Experimental results show that H2C reduces catastrophic forgetting of DL baselines by 22.71% on average in a task-free manner, without relying on manually informed distributional shifts. The implementation is available at https://github.com/BIT-Jack/H2C-lifelong.

cs.AI↗
arXiv2025

Multi-objective Large Language Model Alignment with Hierarchical Experts

Aligning large language models (LLMs) to simultaneously satisfy multiple objectives remains a significant challenge, especially given the diverse and often conflicting nature of human preferences. Existing alignment methods struggle to balance trade-offs effectively, often requiring costly retraining or yielding suboptimal results across the Pareto frontier of preferences. In this paper, we introduce \textit{HoE}(Hierarchical Mixture-of-Experts), a \textit{lightweight}, \textit{parameter-efficient}, and \textit{plug-and-play} approach that eliminates the need for model training, while enabling LLMs to adapt across the entire Pareto frontier and accommodate diverse user preferences. In particular, \textit{HoE} consists of three hierarchical components: LoRA Experts, Router Experts and Preference Routing, reaching optimal Pareto frontiers and achieving a trade-off between parameter size, training cost, and performance. We evaluate \textit{HoE} across various tasks on 14 objectives and 200 different preferences among 6 benchmarks, demonstrating superior performance over 15 recent baselines. Code is available in the supplementary materials.

cs.CL↗
arXiv2025

Knowledge Fusion of Large Language Models Via Modular SkillPacks

Cross-capability transfer is a key challenge in large language model (LLM) research, with applications in multi-task integration, model compression, and continual learning. Recent works like FuseLLM and FuseChat have demonstrated the potential of transferring multiple model capabilities to lightweight models, enhancing adaptability and efficiency, which motivates our investigation into more efficient cross-capability transfer methods. However, existing approaches primarily focus on small, homogeneous models, limiting their applicability. For large, heterogeneous models, knowledge distillation with full-parameter fine-tuning often overlooks the student model's intrinsic capacity and risks catastrophic forgetting, while PEFT methods struggle to effectively absorb knowledge from source LLMs. To address these issues, we introduce GraftLLM, a novel method that stores source model capabilities in a target model with SkillPack format. This approach preserves general capabilities, reduces parameter conflicts, and supports forget-free continual learning and model fusion. We employ a module-aware adaptive compression strategy to compress parameter updates, ensuring efficient storage while maintaining task-specific knowledge. The resulting SkillPack serves as a compact and transferable knowledge carrier, ideal for heterogeneous model fusion and continual learning. Experiments across various scenarios demonstrate that GraftLLM outperforms existing techniques in knowledge transfer, knowledge fusion, and forget-free learning, providing a scalable and efficient solution for cross-capability transfer. The code is publicly available at: https://github.com/duguodong7/GraftLLM.

cs.AI↗
arXiv2025

Neural Parameter Search for Slimmer Fine-Tuned Models and Better Transfer

Foundation models and their checkpoints have significantly advanced deep learning, boosting performance across various applications. However, fine-tuned models often struggle outside their specific domains and exhibit considerable redundancy. Recent studies suggest that combining a pruned fine-tuned model with the original pre-trained model can mitigate forgetting, reduce interference when merging model parameters across tasks, and improve compression efficiency. In this context, developing an effective pruning strategy for fine-tuned models is crucial. Leveraging the advantages of the task vector mechanism, we preprocess fine-tuned models by calculating the differences between them and the original model. Recognizing that different task vector subspaces contribute variably to model performance, we introduce a novel method called Neural Parameter Search (NPS-Pruning) for slimming down fine-tuned models. This method enhances pruning efficiency by searching through neural parameters of task vectors within low-rank subspaces. Our method has three key applications: enhancing knowledge transfer through pairwise model interpolation, facilitating effective knowledge fusion via model merging, and enabling the deployment of compressed models that retain near-original performance while significantly reducing storage costs. Extensive experiments across vision, NLP, and multi-modal benchmarks demonstrate the effectiveness and robustness of our approach, resulting in substantial performance gains. The code is publicly available at: https://github.com/duguodong7/NPS-Pruning.

cs.LG↗
arXiv2024

Enhancing Accuracy and Feature Insights in Hydration Free Energy Predictions for Small Molecules with Machine Learning

The accurate prediction of solvation free energy is of significant importance as it governs the behavior of solutes in solution. In this work, we apply a variety of machine learning techniques to predict and analyze the alchemical free energy of small molecules. Our methodology incorporates an ensemble of machine learning models with feature processing using the K-nearest neighbors algorithm. Two training strategies are explored: one based on experimental data, and the other based on the offset between molecular dynamics (MD) simulations and experimental measurements. The latter approach yields a substantial improvement in predictive accuracy, achieving a mean unsigned error (MUE) of 0.64 kcal/mol. Feature analysis identifies molecular geometry and topology as the most critical factors in predicting alchemical free energy, supporting the established theory that surface tension is a key determinant. Furthermore, the feature analysis of offset results highlights the relevance of charge distribution within the system, which correlates with the inaccuracies in force fields employed in MD simulations and may provide guidance for improving force field designs. These results suggest that machine learning approaches can effectively capture the complex features governing solvation free energy, offering novel pathways for enhancing predictive accuracy.

physics.chem-ph↗
arXiv2024

Parameter Competition Balancing for Model Merging

While fine-tuning pretrained models has become common practice, these models often underperform outside their specific domains. Recently developed model merging techniques enable the direct integration of multiple models, each fine-tuned for distinct tasks, into a single model. This strategy promotes multitasking capabilities without requiring retraining on the original datasets. However, existing methods fall short in addressing potential conflicts and complex correlations between tasks, especially in parameter-level adjustments, posing a challenge in effectively balancing parameter competition across various tasks. This paper introduces an innovative technique named PCB-Merging (Parameter Competition Balancing), a lightweight and training-free technique that adjusts the coefficients of each parameter for effective model merging. PCB-Merging employs intra-balancing to gauge parameter significance within individual tasks and inter-balancing to assess parameter similarities across different tasks. Parameters with low importance scores are dropped, and the remaining ones are rescaled to form the final merged model. We assessed our approach in diverse merging scenarios, including cross-task, cross-domain, and cross-training configurations, as well as out-of-domain generalization. The experimental results reveal that our approach achieves substantial performance enhancements across multiple modalities, domains, model sizes, number of tasks, fine-tuning forms, and large language models, outperforming existing model merging methods. The code is publicly available at: \url{https://github.com/duguodong7/pcb-merging}.

cs.CV↗
arXiv2024

Evolutionary Neural Architecture Search for 3D Point Cloud Analysis

Neural architecture search (NAS) automates neural network design by using optimization algorithms to navigate architecture spaces, reducing the burden of manual architecture design. While NAS has achieved success, applying it to emerging domains, such as analyzing unstructured 3D point clouds, remains underexplored due to the data lying in non-Euclidean spaces, unlike images. This paper presents Success-History-based Self-adaptive Differential Evolution with a Joint Point Interaction Dimension Search (SHSADE-PIDS), an evolutionary NAS framework that encodes discrete deep neural network architectures to continuous spaces and performs searches in the continuous spaces for efficient point cloud neural architectures. Comprehensive experiments on challenging 3D segmentation and classification benchmarks demonstrate SHSADE-PIDS's capabilities. It discovered highly efficient architectures with higher accuracy, significantly advancing prior NAS techniques. For segmentation on SemanticKITTI, SHSADE-PIDS attained 64.51% mean IoU using only 0.55M parameters and 4.5GMACs, reducing overhead by over 22-26X versus other top methods. For ModelNet40 classification, it achieved 93.4% accuracy with just 1.31M parameters, surpassing larger models. SHSADE-PIDS provided valuable insights into bridging evolutionary algorithms with neural architecture optimization, particularly for emerging frontiers like point cloud learning.

cs.CV↗
arXiv2024

Impacts of Darwinian Evolution on Pre-trained Deep Neural Networks

Darwinian evolution of the biological brain is documented through multiple lines of evidence, although the modes of evolutionary changes remain unclear. Drawing inspiration from the evolved neural systems (e.g., visual cortex), deep learning models have demonstrated superior performance in visual tasks, among others. While the success of training deep neural networks has been relying on back-propagation (BP) and its variants to learn representations from data, BP does not incorporate the evolutionary processes that govern biological neural systems. This work proposes a neural network optimization framework based on evolutionary theory. Specifically, BP-trained deep neural networks for visual recognition tasks obtained from the ending epochs are considered the primordial ancestors (initial population). Subsequently, the population evolved with differential evolution. Extensive experiments are carried out to examine the relationships between Darwinian evolution and neural network optimization, including the correspondence between datasets, environment, models, and living species. The empirical results show that the proposed framework has positive impacts on the network, with reduced over-fitting and an order of magnitude lower time complexity compared to BP. Moreover, the experiments show that the proposed framework performs well on deep neural networks and big datasets.

cs.NE↗
arXiv2024

Knowledge Fusion By Evolving Weights of Language Models

Fine-tuning pre-trained language models, particularly large language models, demands extensive computing resources and can result in varying performance outcomes across different domains and datasets. This paper examines the approach of integrating multiple models from diverse training scenarios into a unified model. This unified model excels across various data domains and exhibits the ability to generalize well on out-of-domain data. We propose a knowledge fusion method named Evolver, inspired by evolutionary algorithms, which does not need further training or additional training data. Specifically, our method involves aggregating the weights of different language models into a population and subsequently generating offspring models through mutation and crossover operations. These offspring models are then evaluated against their parents, allowing for the preservation of those models that show enhanced performance on development datasets. Importantly, our model evolving strategy can be seamlessly integrated with existing model merging frameworks, offering a versatile tool for model enhancement. Experimental results on mainstream language models (i.e., encoder-only, decoder-only, encoder-decoder) reveal that Evolver outperforms previous state-of-the-art models by large margins. The code is publicly available at {https://github.com/duguodong7/model-evolution}.

cs.CL↗
arXiv2024

CADE: Cosine Annealing Differential Evolution for Spiking Neural Network

Spiking neural networks (SNNs) have gained prominence for their potential in neuromorphic computing and energy-efficient artificial intelligence, yet optimizing them remains a formidable challenge for gradient-based methods due to their discrete, spike-based computation. This paper attempts to tackle the challenges by introducing Cosine Annealing Differential Evolution (CADE), designed to modulate the mutation factor (F) and crossover rate (CR) of differential evolution (DE) for the SNN model, i.e., Spiking Element Wise (SEW) ResNet. Extensive empirical evaluations were conducted to analyze CADE. CADE showed a balance in exploring and exploiting the search space, resulting in accelerated convergence and improved accuracy compared to existing gradient-based and DE-based methods. Moreover, an initialization method based on a transfer learning setting was developed, pretraining on a source dataset (i.e., CIFAR-10) and fine-tuning the target dataset (i.e., CIFAR-100), to improve population diversity. It was found to further enhance CADE for SNN. Remarkably, CADE elevates the performance of the highest accuracy SEW model by an additional 0.52 percentage points, underscoring its effectiveness in fine-tuning and enhancing SNNs. These findings emphasize the pivotal role of a scheduler for F and CR adjustment, especially for DE-based SNN. Source Code on Github: https://github.com/Tank-Jiang/CADE4SNN.

cs.NE↗