PyTorchKR Papers

최신 논문

총 139,336편
태그·기간 등 상세 필터

카테고리 탐색

arXiv:2503.12811공식 코드 · Papers with Code

A Multi-Power Law for Loss Curve Prediction Across Learning Rate Schedules

Training large models is both resource-intensive and time-consuming, making it crucial to understand the quantitative relationship between model performance and hyperparameters. In this paper, we present an empirical law that describes how the pretraining loss of large language models evolves under different learning rate schedules, such as constant, cosine, and step decay schedules. Our proposed law takes a multi-power form, combining a power law based on the sum of learning rates and additional power laws to account for a loss reduction effect induced by learning rate decay. We extensively validate this law on various model sizes and architectures, and demonstrate that after fitting on a few learning rate schedules, the law accurately predicts the loss curves for unseen schedules of different shapes and horizons. Moreover, by minimizing the predicted final pretraining loss across learning rate schedules, we are able to find a schedule that outperforms the widely used cosine learning rate schedule. Interestingly, this automatically discovered schedule bears some resemblance to the recently proposed Warmup-Stable-Decay (WSD) schedule (Hu et al, 2024) but achieves a slightly lower final loss. We believe these results could offer valuable insights for understanding the dynamics of pretraining and designing learning rate schedules to improve efficiency.

arXiv:2503.12734공식 코드 · Papers with Code

In-Context Linear Regression Demystified: Training Dynamics and Mechanistic Interpretability of Multi-Head Softmax Attention

We study how multi-head softmax attention models are trained to perform in-context learning on linear data. Through extensive empirical experiments and rigorous theoretical analysis, we demystify the emergence of elegant attention patterns: a diagonal and homogeneous pattern in the key-query (KQ) weights, and a last-entry-only and zero-sum pattern in the output-value (OV) weights. Remarkably, these patterns consistently appear from gradient-based training starting from random initialization. Our analysis reveals that such emergent structures enable multi-head attention to approximately implement a debiased gradient descent predictor -- one that outperforms single-head attention and nearly achieves Bayesian optimality up to proportional factor. Furthermore, compared to linear transformers, the softmax attention readily generalizes to sequences longer than those seen during training. We also extend our study to scenarios with non-isotropic covariates and multi-task linear regression. In the former, multi-head attention learns to implement a form of pre-conditioned gradient descent. In the latter, we uncover an intriguing regime where the interplay between head number and task number triggers a superposition phenomenon that efficiently resolves multi-task in-context learning. Our results reveal that in-context learning ability emerges from the trained transformer as an aggregated effect of its architecture and the underlying data distribution, paving the way for deeper understanding and broader applications of in-context learning.

arXiv:2503.08684공식 코드 · Papers with Code

Perplexity Trap: PLM-Based Retrievers Overrate Low Perplexity Documents

Previous studies have found that PLM-based retrieval models exhibit a preference for LLM-generated content, assigning higher relevance scores to these documents even when their semantic quality is comparable to human-written ones. This phenomenon, known as source bias, threatens the sustainable development of the information access ecosystem. However, the underlying causes of source bias remain unexplored. In this paper, we explain the process of information retrieval with a causal graph and discover that PLM-based retrievers learn perplexity features for relevance estimation, causing source bias by ranking the documents with low perplexity higher. Theoretical analysis further reveals that the phenomenon stems from the positive correlation between the gradients of the loss functions in language modeling task and retrieval task. Based on the analysis, a causal-inspired inference-time debiasing method is proposed, called Causal Diagnosis and Correction (CDC). CDC first diagnoses the bias effect of the perplexity and then separates the bias effect from the overall estimated relevance score. Experimental results across three domains demonstrate the superior debiasing effectiveness of CDC, emphasizing the validity of our proposed explanatory framework. Source codes are available at https://github.com/WhyDwelledOnAi/Perplexity-Trap.

arXiv:2503.08643공식 코드 · Papers with Code

Rethinking Diffusion Model in High Dimension

Curse of Dimensionality is an unavoidable challenge in statistical probability models, yet diffusion models seem to overcome this limitation, achieving impressive results in high-dimensional data generation. Diffusion models assume that they can learn the statistical properties of the underlying probability distribution, enabling sampling from this distribution to generate realistic samples. But is this really how they work? To address this question, this paper conducts a detailed analysis of the objective function and inference methods of diffusion models, leading to several important conclusions that help answer the above question: 1) In high-dimensional sparse scenarios, the target of the objective function fitting degrades from a weighted sum of multiple samples to a single sample. 2) The mainstream inference methods can all be represented within a simple unified framework, without requiring statistical concepts such as Markov chains and SDE, while aligning with the degraded objective function. 3) Guided by this simple framework, more efficient inference methods can be discovered.

arXiv:2502.20729공식 코드 · Papers with Code

NeuroMorse: A Temporally Structured Dataset For Neuromorphic Computing

Neuromorphic engineering aims to advance computing by mimicking the brain's efficient processing, where data is encoded as asynchronous temporal events. This eliminates the need for a synchronisation clock and minimises power consumption when no data is present. However, many benchmarks for neuromorphic algorithms primarily focus on spatial features, neglecting the temporal dynamics that are inherent to most sequence-based tasks. This gap may lead to evaluations that fail to fully capture the unique strengths and characteristics of neuromorphic systems. In this paper, we present NeuroMorse, a temporally structured dataset designed for benchmarking neuromorphic learning systems. NeuroMorse converts the top 50 words in the English language into temporal Morse code spike sequences. Despite using only two input spike channels for Morse dots and dashes, complex information is encoded through temporal patterns in the data. The proposed benchmark contains feature hierarchy at multiple temporal scales that test the capacity of neuromorphic algorithms to decompose input patterns into spatial and temporal hierarchies. We demonstrate that our training set is challenging to categorise using a linear classifier and that identifying keywords in the test set is difficult using conventional methods. The NeuroMorse dataset is available at Zenodo, with our accompanying code on GitHub at https://github.com/Ben-E-Walters/NeuroMorse.

arXiv:2502.00213공식 코드 · Papers with Code

Understanding Why Adam Outperforms SGD: Gradient Heterogeneity in Transformers

Transformers are challenging to optimize with SGD and typically require adaptive optimizers such as Adam. However, the reasons behind the superior performance of Adam over SGD remain unclear. In this study, we investigate the optimization of transformers by focusing on gradient heterogeneity, defined as the disparity in gradient norms among parameters. Our analysis shows that gradient heterogeneity hinders gradient-based optimization, including SGD, while sign-based optimization, a simplified variant of Adam, is less affected. We further examine gradient heterogeneity in transformers and show that it is influenced by the placement of layer normalization. Experimental results from fine-tuning transformers in both NLP and vision domains validate our theoretical analyses. This study provides insights into the optimization challenges of transformers and offers guidance for designing future optimization algorithms. Code is available at https://github.com/tom4649/gradient-heterogeneity.

arXiv:2502.00074공식 코드 · Papers with Code

SpikingRTNH: Spiking Neural Network for 4D Radar Object Detection

Recently, 4D Radar has emerged as a crucial sensor for 3D object detection in autonomous vehicles, offering both stable perception in adverse weather and high-density point clouds for object shape recognition. However, processing such high-density data demands substantial computational resources and energy consumption. We propose SpikingRTNH, the first spiking neural network (SNN) for 3D object detection using 4D Radar data. By replacing conventional ReLU activation functions with leaky integrate-and-fire (LIF) spiking neurons, SpikingRTNH achieves significant energy efficiency gains. Furthermore, inspired by human cognitive processes, we introduce biological top-down inference (BTI), which processes point clouds sequentially from higher to lower densities. This approach effectively utilizes points with lower noise and higher importance for detection. Experiments on K-Radar dataset demonstrate that SpikingRTNH with BTI significantly reduces energy consumption by 78% while achieving comparable detection performance to its ANN counterpart (51.1% AP 3D, 57.0% AP BEV). These results establish the viability of SNNs for energy-efficient 4D Radar-based object detection in autonomous driving systems. All codes are available at https://github.com/kaist-avelab/k-radar.

arXiv:2503.21717공식 코드 · Papers with Code

CLAIMCHECK: How Grounded are LLM Critiques of Scientific Papers?

A core part of scientific peer review involves providing expert critiques that directly assess the scientific claims a paper makes. While it is now possible to automatically generate plausible (if generic) reviews, ensuring that these reviews are sound and grounded in the papers' claims remains challenging. To facilitate LLM benchmarking on these challenges, we introduce CLAIMCHECK, an annotated dataset of NeurIPS 2023 and 2024 submissions and reviews mined from OpenReview. CLAIMCHECK is richly annotated by ML experts for weakness statements in the reviews and the paper claims that they dispute, as well as fine-grained labels of the validity, objectivity, and type of the identified weaknesses. We benchmark several LLMs on three claim-centric tasks supported by CLAIMCHECK, requiring models to (1) associate weaknesses with the claims they dispute, (2) predict fine-grained labels for weaknesses and rewrite the weaknesses to enhance their specificity, and (3) verify a paper's claims with grounded reasoning. Our experiments reveal that cutting-edge LLMs, while capable of predicting weakness labels in (2), continue to underperform relative to human experts on all other tasks.

arXiv:2503.21714공식 코드 · Papers with Code

As easy as PIE: understanding when pruning causes language models to disagree

Language Model (LM) pruning compresses the model by removing weights, nodes, or other parts of its architecture. Typically, pruning focuses on the resulting efficiency gains at the cost of effectiveness. However, when looking at how individual data points are affected by pruning, it turns out that a particular subset of data points always bears most of the brunt (in terms of reduced accuracy) when pruning, but this effect goes unnoticed when reporting the mean accuracy of all data points. These data points are called PIEs and have been studied in image processing, but not in NLP. In a study of various NLP datasets, pruning methods, and levels of compression, we find that PIEs impact inference quality considerably, regardless of class frequency, and that BERT is more prone to this than BiLSTM. We also find that PIEs contain a high amount of data points that have the largest influence on how well the model generalises to unseen data. This means that when pruning, with seemingly moderate loss to accuracy across all data points, we in fact hurt tremendously those data points that matter the most. We trace what makes PIEs both hard and impactful to inference to their overall longer and more semantically complex text. These findings are novel and contribute to understanding how LMs are affected by pruning. The code is available at: https://github.com/pietrotrope/AsEasyAsPIE

arXiv:2503.21614공식 코드 · Papers with Code

A Survey of Efficient Reasoning for Large Reasoning Models: Language, Multimodality, and Beyond

Recent Large Reasoning Models (LRMs), such as DeepSeek-R1 and OpenAI o1, have demonstrated strong performance gains by scaling up the length of Chain-of-Thought (CoT) reasoning during inference. However, a growing concern lies in their tendency to produce excessively long reasoning traces, which are often filled with redundant content (e.g., repeated definitions), over-analysis of simple problems, and superficial exploration of multiple reasoning paths for harder tasks. This inefficiency introduces significant challenges for training, inference, and real-world deployment (e.g., in agent-based systems), where token economy is critical. In this survey, we provide a comprehensive overview of recent efforts aimed at improving reasoning efficiency in LRMs, with a particular focus on the unique challenges that arise in this new paradigm. We identify common patterns of inefficiency, examine methods proposed across the LRM lifecycle, i.e., from pretraining to inference, and discuss promising future directions for research. To support ongoing development, we also maintain a real-time GitHub repository tracking recent progress in the field. We hope this survey serves as a foundation for further exploration and inspires innovation in this rapidly evolving area.

arXiv:2503.21513공식 코드 · Papers with Code

Datasets for Depression Modeling in Social Media: An Overview

Depression is the most common mental health disorder, and its prevalence increased during the COVID-19 pandemic. As one of the most extensively researched psychological conditions, recent research has increasingly focused on leveraging social media data to enhance traditional methods of depression screening. This paper addresses the growing interest in interdisciplinary research on depression, and aims to support early-career researchers by providing a comprehensive and up-to-date list of datasets for analyzing and predicting depression through social media data. We present an overview of datasets published between 2019 and 2024. We also make the comprehensive list of datasets available online as a continuously updated resource, with the hope that it will facilitate further interdisciplinary research into the linguistic expressions of depression on social media.

arXiv:2503.21500공식 코드 · Papers with Code

OpenHuEval: Evaluating Large Language Model on Hungarian Specifics

We introduce OpenHuEval, the first benchmark for LLMs focusing on the Hungarian language and specifics. OpenHuEval is constructed from a vast collection of Hungarian-specific materials sourced from multiple origins. In the construction, we incorporated the latest design principles for evaluating LLMs, such as using real user queries from the internet, emphasizing the assessment of LLMs' generative capabilities, and employing LLM-as-judge to enhance the multidimensionality and accuracy of evaluations. Ultimately, OpenHuEval encompasses eight Hungarian-specific dimensions, featuring five tasks and 3953 questions. Consequently, OpenHuEval provides the comprehensive, in-depth, and scientifically accurate assessment of LLM performance in the context of the Hungarian language and its specifics. We evaluated current mainstream LLMs, including both traditional LLMs and recently developed Large Reasoning Models. The results demonstrate the significant necessity for evaluation and model optimization tailored to the Hungarian language and specifics. We also established the framework for analyzing the thinking processes of LRMs with OpenHuEval, revealing intrinsic patterns and mechanisms of these models in non-English languages, with Hungarian serving as a representative example. We will release OpenHuEval at https://github.com/opendatalab/OpenHuEval .

arXiv:2503.21460공식 코드 · Papers with Code

Large Language Model Agent: A Survey on Methodology, Applications and Challenges

The era of intelligent agents is upon us, driven by revolutionary advancements in large language models. Large Language Model (LLM) agents, with goal-driven behaviors and dynamic adaptation capabilities, potentially represent a critical pathway toward artificial general intelligence. This survey systematically deconstructs LLM agent systems through a methodology-centered taxonomy, linking architectural foundations, collaboration mechanisms, and evolutionary pathways. We unify fragmented research threads by revealing fundamental connections between agent design principles and their emergent behaviors in complex environments. Our work provides a unified architectural perspective, examining how agents are constructed, how they collaborate, and how they evolve over time, while also addressing evaluation methodologies, tool applications, practical challenges, and diverse application domains. By surveying the latest developments in this rapidly evolving field, we offer researchers a structured taxonomy for understanding LLM agents and identify promising directions for future research. The collection is available at https://github.com/luo-junyu/Awesome-Agent-Papers.

arXiv:2503.21380공식 코드 · Papers with Code

Challenging the Boundaries of Reasoning: An Olympiad-Level Math Benchmark for Large Language Models

In recent years, the rapid development of large reasoning models has resulted in the saturation of existing benchmarks for evaluating mathematical reasoning, highlighting the urgent need for more challenging and rigorous evaluation frameworks. To address this gap, we introduce OlymMATH, a novel Olympiad-level mathematical benchmark, designed to rigorously test the complex reasoning capabilities of LLMs. OlymMATH features 200 meticulously curated problems, each manually verified and available in parallel English and Chinese versions. The problems are systematically organized into two distinct difficulty tiers: (1) AIME-level problems (easy) that establish a baseline for mathematical reasoning assessment, and (2) significantly more challenging problems (hard) designed to push the boundaries of current state-of-the-art models. In our benchmark, these problems span four core mathematical fields, each including a verifiable numerical solution to enable objective, rule-based evaluation. Empirical results underscore the significant challenge presented by OlymMATH, with state-of-the-art models including DeepSeek-R1, OpenAI's o3-mini and Gemini 2.5 Pro Exp demonstrating notably limited accuracy on the hard subset. Furthermore, the benchmark facilitates comprehensive bilingual assessment of mathematical reasoning abilities-a critical dimension that remains largely unaddressed in mainstream mathematical reasoning benchmarks. We release the benchmark, evaluation code, detailed results and a data visualization tool at https://github.com/RUCAIBox/OlymMATH.

arXiv:2503.21295공식 코드 · Papers with Code

R-PRM: Reasoning-Driven Process Reward Modeling

Large language models (LLMs) inevitably make mistakes when performing step-by-step mathematical reasoning. Process Reward Models (PRMs) have emerged as a promising solution by evaluating each reasoning step. However, existing PRMs typically output evaluation scores directly, limiting both learning efficiency and evaluation accuracy, which is further exacerbated by the scarcity of annotated data. To address these issues, we propose Reasoning-Driven Process Reward Modeling (R-PRM). First, we leverage stronger LLMs to generate seed data from limited annotations, effectively bootstrapping our model's reasoning capabilities and enabling comprehensive step-by-step evaluation. Second, we further enhance performance through preference optimization, without requiring additional annotated data. Third, we introduce inference-time scaling to fully harness the model's reasoning potential. Extensive experiments demonstrate R-PRM's effectiveness: on ProcessBench and PRMBench, it surpasses strong baselines by 11.9 and 8.5 points in F1 scores, respectively. When applied to guide mathematical reasoning, R-PRM achieves consistent accuracy improvements of over 8.5 points across six challenging datasets. Further analysis reveals that R-PRM exhibits more comprehensive evaluation and stronger generalization capabilities, thereby highlighting its significant potential.

arXiv:2503.21261공식 코드 · Papers with Code

HOT: Hadamard-based Optimized Training

It has become increasingly important to optimize backpropagation to reduce memory usage and computational overhead. Achieving this goal is highly challenging, as multiple objectives must be considered jointly while maintaining training quality. In this paper, we focus on matrix multiplication, which accounts for the largest portion of training costs, and analyze its backpropagation in detail to identify lightweight techniques that offer the best benefits. Based on this analysis, we introduce a novel method, Hadamard-based Optimized Training (HOT). In this approach, we apply Hadamard-based optimizations, such as Hadamard quantization and Hadamard low-rank approximation, selectively and with awareness of the suitability of each optimization for different backward paths. Additionally, we introduce two enhancements: activation buffer compression and layer-wise quantizer selection. Our extensive analysis shows that HOT achieves up to 75% memory savings and a 2.6 times acceleration on real GPUs, with negligible accuracy loss compared to FP32 precision.

arXiv:2503.21105공식 코드 · Papers with Code

AugWard: Augmentation-Aware Representation Learning for Accurate Graph Classification

How can we accurately classify graphs? Graph classification is a pivotal task in data mining with applications in social network analysis, web analysis, drug discovery, molecular property prediction, etc. Graph neural networks have achieved the state-of-the-art performance in graph classification, but they consistently struggle with overfitting. To mitigate overfitting, researchers have introduced various representation learning methods utilizing graph augmentation. However, existing methods rely on simplistic use of graph augmentation, which loses augmentation-induced differences and limits the expressiveness of representations. In this paper, we propose AugWard (Augmentation-Aware Training with Graph Distance and Consistency Regularization), a novel graph representation learning framework that carefully considers the diversity introduced by graph augmentation. AugWard applies augmentation-aware training to predict the graph distance between the augmented graph and its original one, aligning the representation difference directly with graph distance at both feature and structure levels. Furthermore, AugWard employs consistency regularization to encourage the classifier to handle richer representations. Experimental results show that AugWard gives the state-of-the-art performance in supervised, semi-supervised graph classification, and transfer learning.

arXiv:2503.21047공식 코드 · Papers with Code

World Model Agents with Change-Based Intrinsic Motivation

Sparse reward environments pose a significant challenge for reinforcement learning due to the scarcity of feedback. Intrinsic motivation and transfer learning have emerged as promising strategies to address this issue. Change Based Exploration Transfer (CBET), a technique that combines these two approaches for model-free algorithms, has shown potential in addressing sparse feedback but its effectiveness with modern algorithms remains understudied. This paper provides an adaptation of CBET for world model algorithms like DreamerV3 and compares the performance of DreamerV3 and IMPALA agents, both with and without CBET, in the sparse reward environments of Crafter and Minigrid. Our tabula rasa results highlight the possibility of CBET improving DreamerV3's returns in Crafter but the algorithm attains a suboptimal policy in Minigrid with CBET further reducing returns. In the same vein, our transfer learning experiments show that pre-training DreamerV3 with intrinsic rewards does not immediately lead to a policy that maximizes extrinsic rewards in Minigrid. Overall, our results suggest that CBET provides a positive impact on DreamerV3 in more complex environments like Crafter but may be detrimental in environments like Minigrid. In the latter case, the behaviours promoted by CBET in DreamerV3 may not align with the task objectives of the environment, leading to reduced returns and suboptimal policies.

arXiv:2503.21023공식 코드 · Papers with Code

Data Mixture Optimization: A Multi-fidelity Multi-scale Bayesian Framework

Careful curation of data sources can significantly improve the performance of LLM pre-training, but predominant approaches rely heavily on intuition or costly trial-and-error, making them difficult to generalize across different data domains and downstream tasks. Although scaling laws can provide a principled and general approach for data curation, standard deterministic extrapolation from small-scale experiments to larger scales requires strong assumptions on the reliability of such extrapolation, whose brittleness has been highlighted in prior works. In this paper, we introduce a $\textit{probabilistic extrapolation framework}$ for data mixture optimization that avoids rigid assumptions and explicitly models the uncertainty in performance across decision variables. We formulate data curation as a sequential decision-making problem$\unicode{x2013}$multi-fidelity, multi-scale Bayesian optimization$\unicode{x2013}$where $\{$data mixtures, model scale, training steps$\}$ are adaptively selected to balance training cost and potential information gain. Our framework naturally gives rise to algorithm prototypes that leverage noisy information from inexpensive experiments to systematically inform costly training decisions. To accelerate methodological progress, we build a simulator based on 472 language model pre-training runs with varying data compositions from the SlimPajama dataset. We observe that even simple kernels and acquisition functions can enable principled decisions across training models from 20M to 1B parameters and achieve $\textbf{2.6x}$ and $\textbf{3.3x}$ speedups compared to multi-fidelity BO and random search baselines. Taken together, our framework underscores potential efficiency gains achievable by developing principled and transferable data mixture optimization methods.

arXiv:2503.20994공식 코드 · Papers with Code

Deep Learning for Forensic Identification of Source

We used contrastive neural networks to learn useful similarity scores between the 144 cartridge casings in the NBIDE dataset, under the common-but-unknown source paradigm. The common-but-unknown source problem is a problem archetype in forensics where the question is whether two objects share a common source (e.g. were two cartridge casings fired from the same firearm). Similarity scores are often used to interpret evidence under this paradigm. We directly compared our results to a state-of-the-art algorithm, Congruent Matching Cells (CMC). When trained on the E3 dataset of 2967 cartridge casings, contrastive learning achieved an ROC AUC of 0.892. The CMC algorithm achieved 0.867. We also conducted an ablation study where we varied the neural network architecture; specifically, the network's width or depth. The ablation study showed that contrastive network performance results are somewhat robust to the network architecture. This work was in part motivated by the use of similarity scores attained via contrastive learning for standard evidence interpretation methods such as score-based likelihood ratios.