arXiv subjects

Xitong Zhang

Publications and source records attributed to Xitong Zhang.

At least 19 recordsLinked to original sources

arXiv2026

How Far Can Sharpness and Complexity Jointly Explain Generalization?

Sharpness and complexity are two central factors in the generalization analysis of deep neural networks. Existing quantitative evaluations of generalization measures have largely focused on individual scalar measures, leaving the joint explanatory power of sharpness and complexity largely unexplored. This work studies how far sharpness and complexity can jointly explain generalization. We use linear regression and introduce a Pareto-based analysis to quantitatively evaluate the joint explanatory power of these two factors. Beyond the existing parameter-level definitions, we further propose realizations of sharpness and complexity that are closer to function space and less dependent on raw parameter representations. We find that function-oriented definitions of these two quantities expand the explanatory scope of the two-factor view beyond what is achieved by existing parameter-level metrics. Overall, our results support the sharpness-complexity perspective as an informative lens for understanding generalization across diverse settings. At the same time, the remaining failures indicate that whether this two-factor view can serve as a complete theory of generalization remains open.

cs.LG↗

arXiv2026

N-Component Free Energy Lattice Boltzmann Method with Reduction Consistency and Global Momentum Conservation

We present a free energy lattice Boltzmann model capable of simulating fluid systems with an arbitrary number of immiscible components in principle. Our method is strictly reduction consistent, ensuring that absent fluid components do not spontaneously nucleate. We introduce a novel discretization of the surface tension force that globally conserves momentum to machine precision, and we enforce reduction consistency through a flux correction that is independent of the mobility. The method is benchmarked with a range of static and dynamic problems, including: liquid lenses, Janus droplets, quaternary phase separation, and six-component layered Poiseuille flow, and we obtain excellent agreement with theoretical predictions throughout. Finally, we demonstrate the applicability of the proposed method through patterned liquid surfaces and microfluidic emulsion droplet generation.

physics.flu-dyn↗

arXiv2026

Understanding the Dynamics of Evaporation-Driven Colloidal Self-Assembly

Complex colloidal cluster morphologies are desirable for the fabrication of advanced materials, such as photonic crystals and meta-materials, and can be formed through evaporation-driven packing. By coupling lattice Boltzmann and discrete element methods, here we elucidate the rich interplay between fluid and particle dynamics during evaporation-driven self-assembly of spherical colloidal particles. We construct a regime diagram for a wide range of evaporation rates, interparticle friction coefficients, and particle numbers, identifying parameter regimes for open, closed, and minimal moment of inertia cluster configurations. Analyzing the competition between capillary, hydrodynamic, normal, and friction forces, we show that interparticle friction can exert a disproportionately strong influence on the final packing outcome despite being considerably smaller in magnitude than other forces at play. Our simulation results further highlight the potential for tuning colloidal cluster configurations via their dynamic trajectories.

cond-mat.soft↗

arXiv2026

Learning to Diagnose and Correct Moral Errors: Beyond Shallow Heuristics in Moral Alignment

Existing approaches to moral value alignment are primarily set out to align LLMs' generation with the distributions of morally appropriate language, which has seen good progress. However, these approaches are often brittle, heavily rely on shallow heuristics, and reduce performance in out-of-the-distribution tasks. In other words, the learning paradigm underlying existing approaches teaches LLMs what morally (in)appropriate language looks like, but not why it is morally (in)appropriate. In this paper, we address this challenge by developing pragmatic inference-driven methods to facilitate LLMs' learning of how to diagnose and correct moral errors, thereby enabling them to generate morally appropriate language. Pragmatic inference is the reasoning process of deriving (implied) meanings -- a famous concept in linguistics. Our methods vary the inference procedures by the inferential load of different moral discourses, rather than modelling their diverse and complex semantic distributions separately. Empirical results demonstrate that our approach improves moral value alignment in LLMs and generalizes effectively across tasks.

cs.CL↗

arXiv2025

Topology Controls the Phase Separation Dynamics of Many Component Fluid Mixtures

Fluid mixtures, ranging from the cellular cytoplasm to synthetic DNA nanostar systems, can spontaneously compartmentalize into many ($N$) coexisting liquid phases through liquid-liquid phase separation. While such systems exhibit a remarkable diversity of spatial organizations, the physical principles governing their non-equilibrium dynamics remain poorly understood. Here, combining simulations and analytical theory, we show that the coarsening dynamics of many component phase separation are fundamentally linked to mathematical coloring problems. For planar phase organization, relevant to synthetic droplet monolayers and simple biological structures, we identify distinct topological constraints for $N=2$, $N=3$, and $N=4$, with no further change for $N>4$, consistent with the four-color theorem. These constraints govern the coarsening dynamics, and, using chromatic graph theory, we derive a theoretical model for $N\geq 3$ that quantitatively captures the diffusive-like coarsening. By contrast, classical theories based solely on Ostwald ripening underestimate the observed dynamics. We further show that tuning interfacial tensions modifies the set of admissible phase arrangements, enabling highly heterogeneous coarsening dynamics across different phases. For unconfined systems with nonplanar phase organization, different coloring constraints apply, with no analogue of the four-color theorem, and coalescence suppression emerges only when the number of phases exceeds $N\gtrsim 7$. More broadly, our work establishes coloring theory as a topological framework for understanding and predicting the dynamics of many component phase-separating fluids.

physics.flu-dyn↗

arXiv2025

On the Convergence of Moral Self-Correction in Large Language Models

Large Language Models (LLMs) are able to improve their responses when instructed to do so, a capability known as self-correction. When instructions provide only a general and abstract goal without specific details about potential issues in the response, LLMs must rely on their internal knowledge to improve response quality, a process referred to as intrinsic self-correction. The empirical success of intrinsic self-correction is evident in various applications, but how and why it is effective remains unknown. Focusing on moral self-correction in LLMs, we reveal a key characteristic of intrinsic self-correction: performance convergence through multi-round interactions; and provide a mechanistic analysis of this convergence behavior. Based on our experimental results and analysis, we uncover the underlying mechanism of convergence: consistently injected self-correction instructions activate moral concepts that reduce model uncertainty, leading to converged performance as the activated moral concepts stabilize over successive rounds. This paper demonstrates the strong potential of moral self-correction by showing that it exhibits a desirable property of converged performance.

cs.CL↗

arXiv2025

From Training to Generalization: Improving Moral Reasoning Through Pragmatic Inference

Although moral reasoning has emerged as a promising research direction for large language models (LLMs), a persistent generalization challenge remains: LLMs often achieve strong performance on training data but struggle to generalize their moral reasoning to unseen test data. From a linguistic perspective, moral reasoning is a pragmatic process in which moral judgments are inferred based on the context of social norms underlying a given moral situation. However, existing approaches overlook this pragmatic nature because of two major bottlenecks: (1) LLMs are primarily skilled in capturing distributional semantics, which differs from the pragmatic nature of moral reasoning; (2) there is currently no effective solution for grounding language in the moral context. In this paper, we develop a pragmatic inference approach that enables LLMs to infer moral judgments for a given moral situation by combining metapragmatic links with Moral Foundations Theory. Specifically, metapragmatic links serve to bridge the gap between distributional semantics and pragmatics, whereas Moral Foundations Theory provides a principled basis for grounding language in moral contexts. Experimental results demonstrate that our approach substantially improves LLMs' generalization in moral reasoning, highlighting the potential of pragmatic inference for future moral reasoning research.

cs.CL↗

arXiv2025

Diagnosing the Performance Trade-off in Moral Alignment: A Case Study on Gender Stereotypes

Moral alignment has emerged as a widely adopted approach for regulating the behavior of pretrained language models (PLMs), typically through fine-tuning on curated datasets. Gender stereotype mitigation is a representational task within the broader application of moral alignment. However, this process often comes at the cost of degraded downstream task performance. Prior studies commonly aim to achieve a performance trade-off by encouraging PLMs to selectively forget only stereotypical knowledge through carefully designed fairness objective, while preserving their language modeling capability (overall forgetting). In this short paper, we investigate whether the performance trade-off can be achieved through the lens of forgetting and the fairness objective. Our analysis shows that the large datasets needed for satisfactory fairness highlight the limitations of current fairness objectives in achieving an effective trade-off: (1) downstream task performance is strongly correlated with overall forgetting; (2) selective forgetting reduces stereotypes, but overall forgetting increases. and (3) general solutions for alleviating forgetting are ineffective at reducing the overall forgetting and fail to improve downstream task performance.

cs.CL↗

arXiv2025

Discourse Heuristics For Paradoxically Moral Self-Correction

Moral self-correction has emerged as a promising approach for aligning the output of Large Language Models (LLMs) with human moral values. However, moral self-correction techniques are subject to two primary paradoxes. First, despite empirical and theoretical evidence to support the effectiveness of self-correction, this LLM capability only operates at a superficial level. Second, while LLMs possess the capability of self-diagnosing immoral aspects of their output, they struggle to identify the cause of this moral inconsistency during their self-correction process. To better understand and address these paradoxes, we analyze the discourse constructions in fine-tuning corpora designed to enhance moral self-correction, uncovering the existence of the heuristics underlying effective constructions. We demonstrate that moral self-correction relies on discourse constructions that reflect heuristic shortcuts, and that the presence of these heuristic shortcuts during self-correction leads to inconsistency when attempting to enhance both self-correction and self-diagnosis capabilities jointly. Based on our findings, we propose a solution to improve moral self-correction by leveraging the heuristics of curated datasets. We also highlight the generalization challenges of this capability, particularly in terms of learning from situated context and model scales.

cs.CL↗

arXiv2025

Prompt Stability Matters: Evaluating and Optimizing Auto-Generated Prompt in General-Purpose Systems

Automatic prompt generation plays a crucial role in enabling general-purpose multi-agent systems to perform diverse tasks autonomously. Existing methods typically evaluate prompts based on their immediate task performance, overlooking the intrinsic qualities that determine their reliability. This outcome-centric view not only limits interpretability but also fails to account for the inherent stochasticity of large language models (LLMs). In this work, we bring attention to prompt stability-the consistency of model responses across repeated executions-as a key factor for building robust and effective prompt generation systems. To quantify this, we propose semantic stability as a criterion for assessing the response consistency of prompts, and fine-tune a LLaMA-based evaluator to measure it automatically across tasks. These components have enabled us to develop the first stability-aware general-purpose prompt generation system that leverages stability feedback to iteratively enhance both prompt quality and system-level performance. Furthermore, we establish a logical chain between prompt stability and task success by analyzing the structural dependencies within our system, proving stability as a necessary condition for effective system-level execution. Empirical results across general and domain-specific tasks demonstrate that our stability-aware framework improves both accuracy and output consistency. By shifting the focus from one-off results to persistent reliability, our work offers a new perspective on prompt design and contributes practical tools for building more trustworthy general-purpose systems.

cs.AI↗

arXiv2025

Modelling droplet-particle interactions on solid surfaces by coupling the lattice Boltzmann and discrete element methods

We develop a three-dimensional numerical scheme for investigating interfacial flows coupled with frictional solid particles. Our approach combines the lattice Boltzmann method (LBM) to model the dynamics of a two-component fluid, and the discrete element method (DEM) to model normal reaction, sliding friction, and rolling friction between solid particles and between particles and solid surfaces. Key to the coupling between the fluid and particle dynamics are the momentum exchange method to transfer hydrodynamic forces between the fluids and particles, a geometric boundary condition to tune particle wettability, and a capillary force model describing surface tension forces between particles and liquid-fluid interfaces. We rigorously validate the contact forces by investigating the dynamics of a particle bouncing off a solid surface and rolling down an inclined plane, the hydrodynamic force by the Segr\`{e}-Silberberg effect, and the capillary force by particle detachment from a liquid-fluid interface. Motivated by the self-cleaning properties of lotus leaves, we apply the method to investigate how drops remove contaminant particles from surfaces. We successfully reproduce scenarios reported experimentally by Naga et al. (Soft Matter (2021) 17(7):1746-1755) by tuning the particle friction. Furthermore, the LBM-DEM approach allows us to systematically explore the effects of particle friction coefficients, drop size, and speed. Our method opens opportunities to study numerous phenomena involving particle dynamics interacting with interfacial flows, including soil erosion, capillary-driven colloidal self-assembly, and how raindrops transport microplastics in the environment. It also makes it possible to control parameters that are difficult to tune independently in experiments, including contact angles, surface tension, and friction coefficients.

physics.flu-dyn↗

arXiv2025

Multiple Equilibria Enables Tunable Wetting of Droplets on Patterned Liquid Surfaces

Patterning solid surfaces with varying wettability is important to manage droplets in microfluidics, heat transfer and printing. Solid surface roughness poses fundamental limitations including contact-line pinning and solid friction. Here, we report an experimental strategy and theoretical design principles for Patterned Liquid Surfaces (PaLS) that combines the controlled wettability from patterning with the ultra-smoothness of a lubricant-infused surface. In contrast to a solid, on PaLS a droplet can be in 10 different wetting states. This richness arises from the adaptation of the liquid lubricants, and can be harnessed to control the apparent contact angle of the droplet over the full range of wettability whilst removing contact-line pinning effects induced by the solid surface. In the limit of thin liquid films, we derive surface-averaged laws for the apparent contact angle for each wetting state, which capture both experimental and simulation data. Our results provide a distinct approach to surface patterning that exploits the interaction of fluids with lubricant-impregnated surfaces.

physics.flu-dyn↗

arXiv2025

Diagnosing Moral Reasoning Acquisition in Language Models: Pragmatics and Generalization

Ensuring that Large Language Models (LLMs) return just responses which adhere to societal values is crucial for their broader application. Prior research has shown that LLMs often fail to perform satisfactorily on tasks requiring moral cognizance, such as ethics-based judgments. While current approaches have focused on fine-tuning LLMs with curated datasets to improve their capabilities on such tasks, choosing the optimal learning paradigm to enhance the ethical responses of LLMs remains an open research debate. In this work, we aim to address this fundamental question: can current learning paradigms enable LLMs to acquire sufficient moral reasoning capabilities? Drawing from distributional semantics theory and the pragmatic nature of moral discourse, our analysis indicates that performance improvements follow a mechanism similar to that of semantic-level tasks, and therefore remain affected by the pragmatic nature of morals latent in discourse, a phenomenon we name the pragmatic dilemma. We conclude that this pragmatic dilemma imposes significant limitations on the generalization ability of current learning paradigms, making it the primary bottleneck for moral reasoning acquisition in LLMs.

cs.CL↗

arXiv2024

Smaller Large Language Models Can Do Moral Self-Correction

Self-correction is one of the most amazing emerging capabilities of Large Language Models (LLMs), enabling LLMs to self-modify an inappropriate output given a natural language feedback which describes the problems of that output. Moral self-correction is a post-hoc approach correcting unethical generations without requiring a gradient update, making it both computationally lightweight and capable of preserving the language modeling ability. Previous works have shown that LLMs can self-debias, and it has been reported that small models, i.e., those with less than 22B parameters, are not capable of moral self-correction. However, there is no direct proof as to why such smaller models fall short of moral self-correction, though previous research hypothesizes that larger models are skilled in following instructions and understanding abstract social norms. In this paper, we empirically validate this hypothesis in the context of social stereotyping, through meticulous prompting. Our experimental results indicate that (i) surprisingly, 3.8B LLMs with proper safety alignment fine-tuning can achieve very good moral self-correction performance, highlighting the significant effects of safety alignment; and (ii) small LLMs are indeed weaker than larger-scale models in terms of comprehending social norms and self-explanation through CoT, but all scales of LLMs show bad self-correction performance given unethical instructions.

cs.CL↗

arXiv2024

Self-correction is Not An Innate Capability in Language Models

Although there has been growing interest in the self-correction capability of Large Language Models (LLMs), there are varying conclusions about its effectiveness. Prior research has largely concentrated on intrinsic self-correction, extrinsic self-correction, particularly the interplay between internal knowledge and external feedback, remains underexplored. In this paper, we aim to comprehensively investigate the underlying mechanism of moral self-correction by addressing a fundamental question: is moral self-correction an innate capability of LLMs? Specifically, we conduct: (1) a behavioral analysis of LLMs' moral sensitivity based on a self-distinguishing task; and (2) a mechanistic analysis of the hidden states to examine how key components of self-correction, such as Chain-of-Thought (CoT) and external feedback, interact to facilitate moral self-correction. Drawing on empirical evidence from both behavioral and mechanistic analyses, we demonstrate that moral self-correction is not an inherent capability of LLMs, as they are neither morally sensitive nor able to effectively incorporate external feedback during the self-correction process.

cs.CL↗

arXiv2024

Optimal Eye Surgeon: Finding Image Priors through Sparse Generators at Initialization

We introduce Optimal Eye Surgeon (OES), a framework for pruning and training deep image generator networks. Typically, untrained deep convolutional networks, which include image sampling operations, serve as effective image priors (Ulyanov et al., 2018). However, they tend to overfit to noise in image restoration tasks due to being overparameterized. OES addresses this by adaptively pruning networks at random initialization to a level of underparameterization. This process effectively captures low-frequency image components even without training, by just masking. When trained to fit noisy images, these pruned subnetworks, which we term Sparse-DIP, resist overfitting to noise. This benefit arises from underparameterization and the regularization effect of masking, constraining them in the manifold of image priors. We demonstrate that subnetworks pruned through OES surpass other leading pruning methods, such as the Lottery Ticket Hypothesis, which is known to be suboptimal for image recovery tasks (Wu et al., 2023). Our extensive experiments demonstrate the transferability of OES-masks and the characteristics of sparse-subnetworks for image generation. Code is available at https://github.com/Avra98/Optimal-Eye-Surgeon.git.

cs.CV↗

arXiv2024

Towards Understanding Task-agnostic Debiasing Through the Lenses of Intrinsic Bias and Forgetfulness

While task-agnostic debiasing provides notable generalizability and reduced reliance on downstream data, its impact on language modeling ability and the risk of relearning social biases from downstream task-specific data remain as the two most significant challenges when debiasing Pretrained Language Models (PLMs). The impact on language modeling ability can be alleviated given a high-quality and long-contextualized debiasing corpus, but there remains a deficiency in understanding the specifics of relearning biases. We empirically ascertain that the effectiveness of task-agnostic debiasing hinges on the quantitative bias level of both the task-specific data used for downstream applications and the debiased model. We empirically show that the lower bound of the bias level of the downstream fine-tuned model can be approximated by the bias level of the debiased model, in most practical cases. To gain more in-depth understanding about how the parameters of PLMs change during fine-tuning due to the forgetting issue of PLMs, we propose a novel framework which can Propagate Socially-fair Debiasing to Downstream Fine-tuning, ProSocialTuning. Our proposed framework can push the fine-tuned model to approach the bias lower bound during downstream fine-tuning, indicating that the ineffectiveness of debiasing can be alleviated by overcoming the forgetting issue through regularizing successfully debiased attention heads based on the PLMs' bias levels from stages of pretraining and debiasing.

cs.CL↗

arXiv2024

On the Intrinsic Self-Correction Capability of LLMs: Uncertainty and Latent Concept

Large Language Models (LLMs) are able to improve their responses when instructed to do so, a capability known as self-correction. When instructions provide only the task's goal without specific details about potential issues in the response, LLMs must rely on their internal knowledge to improve response quality, a process referred to as intrinsic self-correction. The empirical success of intrinsic self-correction is evident in various applications, but how and why it is effective remains unknown. In this paper, we unveil that intrinsic self-correction can be progressively improved, allowing it to approach a converged state. Our findings are verified in: (1) the scenario of multi-round question answering, by comprehensively demonstrating that intrinsic self-correction can progressively introduce performance gains through iterative interactions, ultimately converging to stable performance; and (2) the context of intrinsic self-correction for enhanced morality, in which we provide empirical evidence that iteratively applying instructions reduces model uncertainty towards convergence, which then leads to convergence of both the calibration error and self-correction performance, ultimately resulting in a stable state of intrinsic self-correction. Furthermore, we introduce a mathematical formulation and a simulation task indicating that the latent concepts activated by self-correction instructions drive the reduction of model uncertainty. Based on our experimental results and analysis of the convergence of intrinsic self-correction, we reveal its underlying mechanism: consistent injected instructions reduce model uncertainty which yields converged, improved performance.

cs.CL↗