Recently, building retrieval-augmented generation (RAG) systems to enhance the capability of large language models (LLMs) has become a common practice. Especially in the legal domain, previous judicial decisions play a significant role under the doctrine of stare decisis which emphasizes the importance of making decisions based on (retrieved) prior documents. However, the overall performance of RAG system depends on many components: (1) retrieval corpora, (2) retrieval algorithms, (3) rerankers, (4) LLM backbones, and (5) evaluation metrics. Here we propose LRAGE, an open-source tool for holistic evaluation of RAG systems focusing on the legal domain. LRAGE provides GUI and CLI interfaces to facilitate seamless experiments and investigate how changes in the aforementioned five components affect the overall accuracy. We validated LRAGE using multilingual legal benches including Korean (KBL), English (LegalBench), and Chinese (LawBench) by demonstrating how the overall accuracy changes when varying the five components mentioned above. The source code is available at https://github.com/hoorangyee/LRAGE.

Achieving high-fidelity and temporally smooth 3D human motion generation remains a challenge, particularly within resource-constrained environments. We introduce FlowMotion, a novel method leveraging Conditional Flow Matching (CFM). FlowMotion incorporates a training objective within CFM that focuses on more accurately predicting target motion in 3D human motion generation, resulting in enhanced generation fidelity and temporal smoothness while maintaining the fast synthesis times characteristic of flow-matching-based methods. FlowMotion achieves state-of-the-art jitter performance, achieving the best jitter in the KIT dataset and the second-best jitter in the HumanML3D dataset, and a competitive FID value in both datasets. This combination provides robust and natural motion sequences, off ering a promising equilibrium between generation quality and temporal naturalness. Introduction The synthesis of 3D human body motion has diverse applications across fields such as robotics [1, 2, 3], VR/ AR [4, 5], entertainment [6, 7, 8, 9], and social interaction in virtual 3D spaces [10, 11, 12, 13]. While recent advances in 3D human motion generation are significant, several challenges remain. The inherent complexities of motion generation are exacerbated by the need to incorporate diverse constraints, such as spatial trajectories [14, 15], interactions with surrounding objects [16, 17], or temporal specifications defined by keyframes [18], all aimed at producing lifelike movements. To achieve realistic and context-aware motion synthesis, motion generation techniques frequently leverage data-driven motion capture data. Recent studies in generative models have driven the development of new techniques for synthesizing 3D human body motion. These methods primarily focus on generating realistic movements based on user inputs, especially descriptive text that specifies the intended action. A key advantage of these generative approaches is their ability to produce a diverse range of plausible motion sequences from a single prompt. This allows users to explore multiple interpretations of a desired movement and select the sequence that best aligns with their creative vision.

The exponential growth of Large Multimodal Models (LMMs) has driven advancements in cross-modal reasoning but at significant computational costs. In this work, we focus on visual language models. W e highlight the redundancy and inefficiency in current vision encoders, and seek to construct an adaptive compression method for mul-timodal data. In this work, we characterize a panoply of visual token selection and merging approaches through both benchmarking and qualitative analysis. In particular, we demonstrate that simple cluster-level token aggregation outperforms prior state-of-the-art works in token selection and merging, including merging at the vision encoder level and attention-based approaches. W e underline the redundancy in current vision encoders, and shed light on several puzzling trends regarding principles of visual token selection through cross-modal attention visualizations. This work is a first effort towards more effective encoding and processing of high-dimensional data, and paves the way for more scalable and sustainable multimodal systems.

This paper concerns corpus poisoning attacks in dense information retrieval, where an adversary attempts to compromise the ranking performance of a search algorithm by injecting a small number of maliciously generated documents into the corpus. Our work addresses two limitations in the current literature. First, attacks that perform adversarial gradient-based word substitution search do so in the discrete lexical space, while retrieval itself happens in the continuous embedding space. We thus propose an optimization method that operates in the embedding space directly. Specifically, we train a perturbation model with the objective of maintaining the geometric distance between the original and adversarial document embeddings, while also maximizing the token-level dissimilarity between the original and adversarial documents. Second, it is common for related work to have a strong assumption that the adversary has prior knowledge about the queries. In this paper, we focus on a more challenging variant of the problem where the adversary assumes no prior knowledge about the query distribution (hence, unsupervised). Our core contribution is an adversarial corpus attack that is fast and effective. We present comprehensive experimental results on both in- and out-of-domain datasets, focusing on two related tasks: a top-1 attack and a corpus poisoning attack. We consider attacks under both a white-box and a black-box setting. Notably, our method can generate successful adversarial examples in under two minutes per target document; four times faster compared to the fastest gradient-based word substitution methods in the literature with the same hardware. Furthermore, our adversarial generation method generates text that is more likely to occur under the distribution of natural text (low perplexity), and is therefore more difficult to detect.

Widely shared videos on the internet are often edited. Recently, although Video Large Language Models (Vid-LLMs) have made great progress in general video understanding tasks, their capabilities in video editing understanding (VEU) tasks remain unexplored. To address this gap, in this paper, we introduce VEU-Bench (Video Editing Understanding Benchmark), a comprehensive benchmark that categorizes video editing components across various dimensions, from intra-frame features like shot size to inter-shot attributes such as cut types and transitions. Unlike previous video editing understanding benchmarks that focus mainly on editing element classification, VEU-Bench encompasses 19 fine-grained tasks across three stages: recognition, reasoning, and judging. To enhance the annotation of VEU automatically, we built an annotation pipeline integrated with an ontology-based knowledge base. Through extensive experiments with 11 state-of-the-art Vid-LLMs, our findings reveal that current Vid-LLMs face significant challenges in VEU tasks, with some performing worse than random choice. To alleviate this issue, we develop Oscars, a VEU expert model fine-tuned on the curated VEU-Bench dataset. It outperforms existing open-source Vid-LLMs on VEU-Bench by over 28.3% in accuracy and achieves performance comparable to commercial models like GPT-4o. We also demonstrate that incorporating VEU data significantly enhances the performance of Vid-LLMs on general video understanding benchmarks, with an average improvement of 8.3% across nine reasoning tasks.

The rapidly evolving cloud platforms and the escalating complexity of network traffic demand proper network traffic monitoring and anomaly detection to ensure network security and performance. This paper introduces a large language model (LLM)-based network traffic monitoring and anomaly detection system. In addition to existing models such as autoencoders and decision trees, we harness the power of large language models for processing sequence data from network traffic, which allows us a better capture of underlying complex patterns, as well as slight fluctuations in the dataset. We show for a given detection task, the need for a hybrid model that incorporates the attention mechanism of the transformer architecture into a supervised learning framework in order to achieve better accuracy. A pre-trained large language model analyzes and predicts the probable network traffic, and an anomaly detection layer that considers temporality and context is added. Moreover, we present a novel transfer learning-based methodology to enhance the model's effectiveness to quickly adapt to unknown network structures and adversarial conditions without requiring extensive labeled datasets. Actual results show that the designed model outperforms traditional methods in detection accuracy and computational efficiency, effectively identify various network anomalies such as zero-day attacks and traffic congestion pattern, and significantly reduce the false positive rate.

Optimization algorithms are widely employed to tackle complex problems, but designing them manually is often labor-intensive and requires significant expertise. Global placement is a fundamental step in electronic design automation (EDA). While analytical approaches represent the state-of-the-art (SOTA) in global placement, their core optimization algorithms remain heavily dependent on heuristics and customized components, such as initialization strategies, preconditioning methods, and line search techniques. This paper presents an automated framework that leverages large language models (LLM) to evolve optimization algorithms for global placement. We first generate diverse candidate algorithms using LLM through carefully crafted prompts. Then we introduce an LLM-based genetic flow to evolve selected candidate algorithms. The discovered optimization algorithms exhibit substantial performance improvements across many benchmarks. Specifically, Our design-case-specific discovered algorithms achieve average HPWL improvements of \textbf{5.05\%}, \text{5.29\%} and \textbf{8.30\%} on MMS, ISPD2005 and ISPD2019 benchmarks, and up to \textbf{17\%} improvements on individual cases. Additionally, the discovered algorithms demonstrate good generalization ability and are complementary to existing parameter-tuning methods.

However, the method of selecting -1, 0, and 1 to determine the summation and subtraction may not be optimal, as there are only a limited number of combinations of -1, 0, and 1, leading to repetitive computations for the corresponding A entries. Furthermore, when increasing computation parallelism by duplicating the selection unit of the adding and subtracting path, the resource consumption of the selection may exceed that of the TL tables themselves. This is because the multiple reading ports of the on-chip distributed RAM unit can support multiple accesses to the TL tables, requiring only additional buffers for addressing. The supportive ablation study will be presented in the next subsection. Algorithm 1: TL-based Ternary Matmul Input: A: Input activation stream (shape r M sr N s); W idx = Offline preprocess(W): Offline-preprocessed weight indices (shape r N {p T G qsrK s); Output: O: Output activation stream (shape r M sr K s); Initialize:; TL TABLEr N sr 3G s Ð 0; // Table for all signed combinations A BLOCK r T ˆ G s Ð 0; // Activation buffer O BLOCK r K s Ð 0; // Output vector accumulator for i Ð 0 to M 1 do for j Ð 0 to N 1 step T ˆ G do // Load activation block for p Ð 0 to T ˆ G 1 do A BLOCK r ps Ð A.read(); end for // Set up values of TL_TABLE for t Ð 0 to T 1 do val 1...G Ð A BLOCK r t ˆ G: p t ` 1 q ˆ G 1s; TL T ABLE set up pval 1...G q; end for // Process hidden dimension for m Ð 0 to K step Q do for n Ð 0 to Q 1 do idx vec Ð B "Y j T ˆG ]ı r m ` n s; for t Ð 0 to N 1 do TL TABLE idx Ð idx vecr t s; O BLOCK r m ` n s Ð O BLOCK rm ` n s ` TL TABLErt sr TL TABLE idx s; end for end for end for end for // Write output for p Ð 0 to K 1 do O .writep C BLOCK r p sq; O BLOCK r ps Ð 0; end for end for Function Offline preprocess(W): return Encode every G value as an index in the matrix and pack every T values as a index vector idx vec; Function TL TABLE set up(val 1...G): return return all 3G add and subtract combination; As described in Algorithm 1 and Figure 1, the TL-based matmul can be divided into two stages: (1) preprocessing the weights into groups sized G, and (2) performing the online ternary matrix multiplication computation. Figure 1: Dataflow and architecture of TL-based ternary matMul (G " 4) In the preprocessing stage, assume that every G " 3 ternary values are packed into a single index for TL table addressing, resulting in 3 G " 3 ˆ 3 ˆ 3 " 27 combinations. The index representation for this packing requires log 2 27 5 bits. Let A P N M ˆ N and W P ternary N ˆK . The preprocessing of the weights involves encoding every group of G " 3 ternary values into a 5-bit packed index.

Clinical document classification is essential for converting unstructured medical texts into standardised ICD-10 diagnoses, yet it faces challenges due to complex medical language, privacy constraints, and limited annotated datasets. Large Language Models (LLMs) offer promising improvements in accuracy and efficiency for this task. This study evaluates the performance and consistency of eight LLMs; four reasoning (Qwen QWQ, Deepseek Reasoner, GPT o3 Mini, Gemini 2.0 Flash Thinking) and four non-reasoning (Llama 3.3, GPT 4o Mini, Gemini 2.0 Flash, Deepseek Chat); in classifying clinical discharge summaries using the MIMIC-IV dataset. Using cTAKES to structure clinical narratives, models were assessed across three experimental runs, with majority voting determining final predictions. Results showed that reasoning models outperformed non-reasoning models in accuracy (71% vs 68%) and F1 score (67% vs 60%), with Gemini 2.0 Flash Thinking achieving the highest accuracy (75%) and F1 score (76%). However, non-reasoning models demonstrated greater stability (91% vs 84% consistency). Performance varied across ICD-10 codes, with reasoning models excelling in complex cases but struggling with abstract categories. Findings indicate a trade-off between accuracy and consistency, suggesting that a hybrid approach could optimise clinical coding. Future research should explore multi-label classification, domain-specific fine-tuning, and ensemble methods to enhance model reliability in real-world applications.

Reinforcement Learning from Human Feedback (RLHF) has become a pivotal paradigm in artificial intelligence to align large models with human preferences. In this paper, we propose a novel statistical framework to simultaneously conduct the online decision-making and statistical inference on the optimal model using human preference data based on dynamic contextual information. Our approach introduces an efficient decision strategy that achieves both the optimal regret bound and the asymptotic distribution of the estimators. A key challenge in RLHF is handling the dependent online human preference outcomes with dynamic contexts. To address this, in the methodological aspect, we propose a two-stage algorithm starting with $\epsilon$-greedy followed by exploitations; in the theoretical aspect, we tailor anti-concentration inequalities and matrix martingale concentration techniques to derive the uniform estimation rate and asymptotic normality of the estimators using dependent samples from both stages. Extensive simulation results demonstrate that our method outperforms state-of-the-art strategies. We apply the proposed framework to analyze the human preference data for ranking large language models on the Massive Multitask Language Understanding dataset, yielding insightful results on the performance of different large language models for medical anatomy knowledge.