PyTorchKR Papers

최신 논문

총 1,975편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2409.17113

Characterizing stable regions in the residual stream of LLMs

We identify stable regions in the residual stream of Transformers, where the model's output remains insensitive to small activation changes, but exhibits high sensitivity at region boundaries. These regions emerge during training and become more defined as training progresses or model size increases. The regions appear to be much larger than previously studied polytopes. Our analysis suggests that these stable regions align with semantic distinctions, where similar prompts cluster within regions, and activations from the same region lead to similar next token predictions. This work provides a promising research direction for understanding the complexity of neural networks, shedding light on training dynamics, and advancing interpretability.

arXiv:2409.16909

Enhancing Temporal Sensitivity and Reasoning for Time-Sensitive Question Answering

Time-Sensitive Question Answering (TSQA) demands the effective utilization of specific temporal contexts, encompassing multiple time-evolving facts, to address time-sensitive questions. This necessitates not only the parsing of temporal information within questions but also the identification and understanding of time-evolving facts to generate accurate answers. However, current large language models still have limited sensitivity to temporal information and their inadequate temporal reasoning capabilities. In this paper, we propose a novel framework that enhances temporal awareness and reasoning through Temporal Information-Aware Embedding and Granular Contrastive Reinforcement Learning. Experimental results on four TSQA datasets demonstrate that our framework significantly outperforms existing LLMs in TSQA tasks, marking a step forward in bridging the performance gap between machine and human temporal understanding and reasoning.

arXiv:2409.16086

Assessing Simplification Levels in Neural Networks: The Impact of Hyperparameter Configurations on Complexity and Sensitivity

This paper presents an experimental study focused on understanding the simplification properties of neural networks under different hyperparameter configurations, specifically investigating the effects on Lempel Ziv complexity and sensitivity. By adjusting key hyperparameters such as activation functions, hidden layers, and learning rate, this study evaluates how these parameters impact the complexity of network outputs and their robustness to input perturbations. The experiments conducted using the MNIST dataset aim to provide insights into the relationships between hyperparameters, complexity, and sensitivity, contributing to a deeper theoretical understanding of these concepts in neural networks.

arXiv:2409.15019

Evaluating Synthetic Activations composed of SAE Latents in GPT-2

Sparse Auto-Encoders (SAEs) are commonly employed in mechanistic interpretability to decompose the residual stream into monosemantic SAE latents. Recent work demonstrates that perturbing a model's activations at an early layer results in a step-function-like change in the model's final layer activations. Furthermore, the model's sensitivity to this perturbation differs between model-generated (real) activations and random activations. In our study, we assess model sensitivity in order to compare real activations to synthetic activations composed of SAE latents. Our findings indicate that synthetic activations closely resemble real activations when we control for the sparsity and cosine similarity of the constituent SAE latents. This suggests that real activations cannot be explained by a simple "bag of SAE latents" lacking internal structure, and instead suggests that SAE latents possess significant geometric and statistical properties. Notably, we observe that our synthetic activations exhibit less pronounced activation plateaus compared to those typically surrounding real activations.

arXiv:2409.14759

VLM's Eye Examination: Instruct and Inspect Visual Competency of Vision Language Models

Vision language models (VLMs) have shown promising reasoning capabilities across various benchmarks; however, our understanding of their visual perception remains limited. In this work, we propose an eye examination process to investigate how a VLM perceives images, specifically focusing on key elements of visual recognition, from primitive color and shape to semantic levels. To this end, we introduce a dataset named LENS to guide a VLM to follow the examination and check its readiness. Once the model is ready, we conduct the examination. Through this examination, we quantify and visualize VLMs' sensitivities to color and shape, and semantic matching. Our findings reveal that VLMs have varying sensitivity to different colors while consistently showing insensitivity to green across different VLMs. Also, we found different shape sensitivity and semantic recognition depending on LLM's capacity despite using the same fixed visual encoder. Our analyses and findings have potential to inspire the design of VLMs and the pre-processing of visual input to VLMs for improving application performance.

arXiv:2409.14456

Scoring rule nets: beyond mean target prediction in multivariate regression

Probabilistic regression models trained with maximum likelihood estimation (MLE), can sometimes overestimate variance to an unacceptable degree. This is mostly problematic in the multivariate domain. While univariate models often optimize the popular Continuous Ranked Probability Score (CRPS), in the multivariate domain, no such alternative to MLE has yet been widely accepted. The Energy Score - the most investigated alternative - notoriously lacks closed-form expressions and sensitivity to the correlation between target variables. In this paper, we propose Conditional CRPS: a multivariate strictly proper scoring rule that extends CRPS. We show that closed-form expressions exist for popular distributions and illustrate their sensitivity to correlation. We then show in a variety of experiments on both synthetic and real data, that Conditional CRPS often outperforms MLE, and produces results comparable to state-of-the-art non-parametric models, such as Distributional Random Forest (DRF).

arXiv:2409.14107

Routing in Sparsely-gated Language Models responds to Context

Language Models (LMs) recently incorporate mixture-of-experts layers consisting of a router and a collection of experts to scale up their parameter count given a fixed computational budget. Building on previous efforts indicating that token-expert assignments are predominantly influenced by token identities and positions, we trace routing decisions of similarity-annotated text pairs to evaluate the context sensitivity of learned token-expert assignments. We observe that routing in encoder layers mainly depends on (semantic) associations, but contextual cues provide an additional layer of refinement. Conversely, routing in decoder layers is more variable and markedly less sensitive to context.

arXiv:2409.13843공식 코드 · Papers with Code

STOP! Benchmarking Large Language Models with Sensitivity Testing on Offensive Progressions

Mitigating explicit and implicit biases in Large Language Models (LLMs) has become a critical focus in the field of natural language processing. However, many current methodologies evaluate scenarios in isolation, without considering the broader context or the spectrum of potential biases within each situation. To address this, we introduce the Sensitivity Testing on Offensive Progressions (STOP) dataset, which includes 450 offensive progressions containing 2,700 unique sentences of varying severity that progressively escalate from less to more explicitly offensive. Covering a broad spectrum of 9 demographics and 46 sub-demographics, STOP ensures inclusivity and comprehensive coverage. We evaluate several leading closed- and open-source models, including GPT-4, Mixtral, and Llama 3. Our findings reveal that even the best-performing models detect bias inconsistently, with success rates ranging from 19.3% to 69.8%. We also demonstrate how aligning models with human judgments on STOP can improve model answer rates on sensitive tasks such as BBQ, StereoSet, and CrowS-Pairs by up to 191%, while maintaining or even improving performance. STOP presents a novel framework for assessing the complex nature of biases in LLMs, which will enable more effective bias mitigation strategies and facilitates the creation of fairer language models.

arXiv:2409.13727공식 코드 · Papers with Code

Classification performance and reproducibility of GPT-4 omni for information extraction from veterinary electronic health records

Large language models (LLMs) can extract information from veterinary electronic health records (EHRs), but performance differences between models, the effect of temperature settings, and the influence of text ambiguity have not been previously evaluated. This study addresses these gaps by comparing the performance of GPT-4 omni (GPT-4o) and GPT-3.5 Turbo under different conditions and investigating the relationship between human interobserver agreement and LLM errors. The LLMs and five humans were tasked with identifying six clinical signs associated with Feline chronic enteropathy in 250 EHRs from a veterinary referral hospital. At temperature 0, the performance of GPT-4o compared to the majority opinion of human respondents, achieved 96.9% sensitivity (interquartile range [IQR] 92.9-99.3%), 97.6% specificity (IQR 96.5-98.5%), 80.7% positive predictive value (IQR 70.8-84.6%), 99.5% negative predictive value (IQR 99.0-99.9%), 84.4% F1 score (IQR 77.3-90.4%), and 96.3% balanced accuracy (IQR 95.0-97.9%). The performance of GPT-4o was significantly better than that of its predecessor, GPT-3.5 Turbo, particularly with respect to sensitivity where GPT-3.5 Turbo only achieved 81.7% (IQR 78.9-84.8%). Adjusting the temperature for GPT-4o did not significantly impact classification performance. GPT-4o demonstrated greater reproducibility than human pairs regardless of temperature, with an average Cohen's kappa of 0.98 (IQR 0.98-0.99) at temperature 0 compared to 0.8 (IQR 0.78-0.81) for humans. Most GPT-4o errors occurred in instances where humans disagreed (35/43 errors, 81.4%), suggesting that these errors were more likely caused by ambiguity of the EHR than explicit model faults. Using GPT-4o to automate information extraction from veterinary EHRs is a viable alternative to manual extraction.

arXiv:2409.12816

Hierarchical Gradient-Based Genetic Sampling for Accurate Prediction of Biological Oscillations

Biological oscillations are periodic changes in various signaling processes crucial for the proper functioning of living organisms. These oscillations are modeled by ordinary differential equations, with coefficient variations leading to diverse periodic behaviors, typically measured by oscillatory frequencies. This paper explores sampling techniques for neural networks to model the relationship between system coefficients and oscillatory frequency. However, the scarcity of oscillations in the vast coefficient space results in many samples exhibiting non-periodic behaviors, and small coefficient changes near oscillation boundaries can significantly alter oscillatory properties. This leads to non-oscillatory bias and boundary sensitivity, making accurate predictions difficult. While existing importance and uncertainty sampling approaches partially mitigate these challenges, they either fail to resolve the sensitivity problem or result in redundant sampling. To address these limitations, we propose the Hierarchical Gradient-based Genetic Sampling (HGGS) framework, which improves the accuracy of neural network predictions for biological oscillations. The first layer, Gradient-based Filtering, extracts sensitive oscillation boundaries and removes redundant non-oscillatory samples, creating a balanced coarse dataset. The second layer, Multigrid Genetic Sampling, utilizes residual information to refine these boundaries and explore new high-residual regions, increasing data diversity for model training. Experimental results demonstrate that HGGS outperforms seven comparative sampling methods across four biological systems, highlighting its effectiveness in enhancing sampling and prediction accuracy.

arXiv:2409.12820공식 코드 · Papers with Code

Machine-learning based high-bandwidth magnetic sensing

Recent years have seen significant growth of quantum technologies, and specifically quantum sensing, both in terms of the capabilities of advanced platforms and their applications. One of the leading platforms in this context is nitrogen-vacancy (NV) color centers in diamond, providing versatile, high-sensitivity, and high-spatial-resolution magnetic sensing. Nevertheless, current schemes for spin resonance magnetic sensing (as applied by NV quantum sensing) suffer from tradeoffs associated with sensitivity, dynamic range, and bandwidth. Here we address this issue, and implement machine learning tools to enhance NV magnetic sensing in terms of the sensitivity/bandwidth tradeoff in large dynamic range scenarios. Our results indicate a potential reduction of required data points by at least a factor of 3, while maintaining the current error level. Our results promote quantum machine learning protocols for sensing applications towards more feasible and efficient quantum technologies.

arXiv:2409.11449공식 코드 · Papers with Code

Evaluation of pretrained language models on music understanding

Music-text multimodal systems have enabled new approaches to Music Information Research (MIR) applications such as audio-to-text and text-to-audio retrieval, text-based song generation, and music captioning. Despite the reported success, little effort has been put into evaluating the musical knowledge of Large Language Models (LLM). In this paper, we demonstrate that LLMs suffer from 1) prompt sensitivity, 2) inability to model negation (e.g. 'rock song without guitar'), and 3) sensitivity towards the presence of specific words. We quantified these properties as a triplet-based accuracy, evaluating the ability to model the relative similarity of labels in a hierarchical ontology. We leveraged the Audioset ontology to generate triplets consisting of an anchor, a positive (relevant) label, and a negative (less relevant) label for the genre and instruments sub-tree. We evaluated the triplet-based musical knowledge for six general-purpose Transformer-based models. The triplets obtained through this methodology required filtering, as some were difficult to judge and therefore relatively uninformative for evaluation purposes. Despite the relatively high accuracy reported, inconsistencies are evident in all six models, suggesting that off-the-shelf LLMs need adaptation to music before use.

arXiv:2409.10673공식 코드 · Papers with Code

A Bayesian Interpretation of Adaptive Low-Rank Adaptation

Motivated by the sensitivity-based importance score of the adaptive low-rank adaptation (AdaLoRA), we utilize more theoretically supported metrics, including the signal-to-noise ratio (SNR), along with the Improved Variational Online Newton (IVON) optimizer, for adaptive parameter budget allocation. The resulting Bayesian counterpart not only has matched or surpassed the performance of using the sensitivity-based importance metric but is also a faster alternative to AdaLoRA with Adam. Our theoretical analysis reveals a significant connection between the two metrics, providing a Bayesian perspective on the efficacy of sensitivity as an importance score. Furthermore, our findings suggest that the magnitude, rather than the variance, is the primary indicator of the importance of parameters.

arXiv:2409.09648

SciDVS: A Scientific Event Camera with 1.7% Temporal Contrast Sensitivity at 0.7 lux

This paper reports a Dynamic Vision Sensor (DVS) event camera that is 6x more sensitive at 14x lower illumination than existing commercial and prototype cameras. Event cameras output a sparse stream of brightness change events. Their high dynamic range (HDR), quick response, and high temporal resolution provide key advantages for scientific applications that involve low lighting conditions and sparse visual events. However, current DVS are hindered by low sensitivity, resulting from shot noise and pixel-to-pixel mismatch. Commercial DVS have a minimum brightness change threshold of >10%. Sensitive prototypes achieved as low as 1%, but required kilo-lux illumination. Our SciDVS prototype fabricated in a 180nm CMOS image sensor process achieves 1.7% sensitivity at chip illumination of 0.7 lx and 18 Hz bandwidth. Novel features of SciDVS are (1) an auto-centering in-pixel preamplifier providing intrascene HDR and increased sensitivity, (2) improved control of bandwidth to limit shot noise, and (3) optional pixel binning, allowing the user to trade spatial resolution for sensitivity.

arXiv:2409.08347

Sensitivity analysis of the perturbed utility stochastic traffic equilibrium

This paper develops a novel sensitivity analysis framework for the perturbed utility route choice (PURC) model and the accompanying stochastic traffic equilibrium model. We provide general results that determine the marginal change in link flows following a marginal change in link costs across the network in the cases of flow-independent and flow-dependent link costs. We derive analytical sensitivity expressions for the Jacobian of the individual optimal PURC flow and equilibrium link flows with respect to link cost parameters under mild differentiability assumptions. Numerical examples illustrate the robustness of our method, demonstrating its use for estimating equilibrium link flows after link cost shifts, identifying critical design parameters, and quantifying uncertainty in performance predictions. The findings have implications for network design, pricing strategies, and policy analysis in transportation planning and economics, providing a bridge between theoretical models and real-world applications.

arXiv:2409.08953공식 코드 · Papers with Code

Pushing the boundaries of event subsampling in event-based video classification using CNNs

Event cameras offer low-power visual sensing capabilities ideal for edge-device applications. However, their high event rate, driven by high temporal details, can be restrictive in terms of bandwidth and computational resources. In edge AI applications, determining the minimum amount of events for specific tasks can allow reducing the event rate to improve bandwidth, memory, and processing efficiency. In this paper, we study the effect of event subsampling on the accuracy of event data classification using convolutional neural network (CNN) models. Surprisingly, across various datasets, the number of events per video can be reduced by an order of magnitude with little drop in accuracy, revealing the extent to which we can push the boundaries in accuracy vs. event rate trade-off. Additionally, we also find that lower classification accuracy in high subsampling rates is not solely attributable to information loss due to the subsampling of the events, but that the training of CNNs can be challenging in highly subsampled scenarios, where the sensitivity to hyperparameters increases. We quantify training instability across multiple event-based classification datasets using a novel metric for evaluating the hyperparameter sensitivity of CNNs in different subsampling settings. Finally, we analyze the weight gradients of the network to gain insight into this instability.

arXiv:2409.08633

Improving Analog Neural Network Robustness: A Noise-Agnostic Approach with Explainable Regularizations

This work tackles the critical challenge of mitigating "hardware noise" in deep analog neural networks, a major obstacle in advancing analog signal processing devices. We propose a comprehensive, hardware-agnostic solution to address both correlated and uncorrelated noise affecting the activation layers of deep neural models. The novelty of our approach lies in its ability to demystify the "black box" nature of noise-resilient networks by revealing the underlying mechanisms that reduce sensitivity to noise. In doing so, we introduce a new explainable regularization framework that harnesses these mechanisms to significantly enhance noise robustness in deep neural architectures.

arXiv:2409.05273

A new method to derive the calculation formula of antenna total isotropic sensitivity

In the OTA test of the antenna spatial performance of communication equipment,the total radiated power TRP and the total isotropic sensitivity TIS are two important characterization parameters,representing the overall transmission and reception performance of the antenna in space.In order to gain a deeper understanding of the physical meaning of these two parameters and the calculation method of TIS,this paper proposes a new method for deriving the TIS calculation formula.This method uses the reciprocity principle of the transmitting antenna and the receiving antenna.By comparing the transmitting process,two corresponding receiving antenna parameters are newly defined in the receiving process analysis.Combining the relationship between the antenna pattern and the receiving intensity,the relationship formula between TIS and equivalent isotropic sensitivity EIS is obtained after a simple deduction.This formula is consistent with the calculation formula given in the reference,but the derivation method in the reference is to use the ideal omnidirectional antenna as a reference,calculate the received power at the receiver input,and combine factors such as the actual antenna gain to derive the TIS calculation formula;the method proposed in this paper is to first derive the expression of the receiving directivity coefficient by definition,introduce the receiving strength parameter,calculate the full-space receiving power at the antenna,and derive the TIS calculation formula.In comparison,the derivation process in this paper calculates the received power from the input end of the receiving antenna,so there is no need to consider the efficiency of the antenna itself.The received power calculation starts directly from the receiving intensity.The entire derivation process is intuitive and easy to understand,avoiding complex concepts and formulas.

arXiv:2409.07538

Un índice discreto sensible a la desigualdad

This paper introduces the Relative Inequality Index at the Maximum (IDRM), a novel and intuitive measure designed to capture inequality within a population, such as income inequality. The index is based on the idea that individuals experience varying levels of inequality depending on their position within the distribution, particularly with respect to those at the top. The key assumption is that for individuals in lower positions, inequalities referenced to the top positions have greater impact on their well-being and the inequality relative to maximum is the most critical. The IDRM fulfills desirable theoretical properties which were used for its evaluation and comparison against widely accepted measures in inequality literature. From this perspective, the IDRM is shown to be as robust as traditional measures and outperforms the Gini and Dalton indices by satisfying eight out of nine key properties, including decomposability across population subgroups. In a comparative analysis using income data from 58 countries and microdata from Mexico, with the Gini, Theil, and Atkinson indices as benchmarks, the IDRM demonstrates superior consistency, sensitivity to inequality, reduced bias in grouped data, and enhanced precision. This index reflects the varying forms of income distribution, showing heightened sensitivity to the magnitude of inequality.

arXiv:2409.03244

Grid-Forming Storage Networks: Analytical Characterization of Damping and Design Insights

The paper presents a theoretical study on small-signal stability and damping in bulk power systems with multiple grid-forming inverter-based storage resources. A detailed analysis is presented, characterizing the impacts of inverter droop gains and storage size on the slower eigenvalues, particularly those concerning inter-area oscillation modes. From these parametric sensitivity studies, a set of necessary conditions are derived that the design of droop gain must satisfy to enhance damping performance. The analytical findings are structured into propositions highlighting potential design considerations for improving system stability. The findings are illustrated via numerical studies on an IEEE 68-bus grid-forming storage network.