Entity-level sentiment classification involves identifying the sentiment polarity linked to specific entities within text. This task poses several challenges: effectively modeling the subtle and complex interactions between entities and their surrounding sentiment expressions; capturing dependencies that may span across sentences; and ensuring consistent sentiment predictions for multiple mentions of the same entity through coreference resolution. Additionally, linguistic phenomena such as negation, ambiguity, and overlapping opinions further complicate the analysis. These complexities make entity-level sentiment classification a difficult problem, especially in real-world, noisy textual data. To address these issues, we propose SpanEIT, a novel framework integrating dynamic span interaction and graph-aware memory mechanisms for enhanced entity-sentiment relational modeling. SpanEIT builds span-based representations for entities and candidate sentiment phrases, employs bidirectional attention for fine-grained interactions, and uses a graph attention network to capture syntactic and co-occurrence relations. A coreference-aware memory module ensures entity-level consistency across documents. Experiments on FSAD, BARU, and IMDB datasets show SpanEIT outperforms state-of-the-art transformer and hybrid baselines in accuracy and F1 scores. Ablation and interpretability analyses validate the effectiveness of our approach, underscoring its potential for fine-grained sentiment analysis in applications like social media monitoring and customer feedback analysis.

Predicting the success of start-up companies, defined as achieving an exit through acquisition or IPO, is a critical problem in entrepreneurship and innovation research. Datasets such as Crunchbase provide both structured information (e.g., funding rounds, industries, investor networks) and unstructured text (e.g., company descriptions), but effectively leveraging this heterogeneous data for prediction remains challenging. Traditional machine learning approaches often rely only on structured features and achieve moderate accuracy, while large language models (LLMs) offer rich reasoning abilities but struggle to adapt directly to domain-specific business data. We present \textbf{CrunchLLM}, a domain-adapted LLM framework for startup success prediction. CrunchLLM integrates structured company attributes with unstructured textual narratives and applies parameter-efficient fine-tuning strategies alongside prompt optimization to specialize foundation models for entrepreneurship data. Our approach achieves accuracy exceeding 80\% on Crunchbase startup success prediction, significantly outperforming traditional classifiers and baseline LLMs. Beyond predictive performance, CrunchLLM provides interpretable reasoning traces that justify its predictions, enhancing transparency and trustworthiness for financial and policy decision makers. This work demonstrates how adapting LLMs with domain-aware fine-tuning and structured--unstructured data fusion can advance predictive modeling of entrepreneurial outcomes. CrunchLLM contributes a methodological framework and a practical tool for data-driven decision making in venture capital and innovation policy.

Smart contracts automate the management of high-value assets, where vulnerabilities can lead to catastrophic financial losses. This challenge is amplified in Large Language Models (LLMs) by two interconnected failures: they operate as unauditable "black boxes" lacking a transparent reasoning process, and consequently, generate code riddled with critical security vulnerabilities. To address both issues, we propose SmartCoder-R1 (based on Qwen2.5-Coder-7B), a novel framework for secure and explainable smart contract generation. It begins with Continual Pre-training (CPT) to specialize the model. We then apply Long Chain-of-Thought Supervised Fine-Tuning (L-CoT SFT) on 7,998 expert-validated reasoning-and-code samples to train the model to emulate human security analysis. Finally, to directly mitigate vulnerabilities, we employ Security-Aware Group Relative Policy Optimization (S-GRPO), a reinforcement learning phase that refines the generation policy by optimizing a weighted reward signal for compilation success, security compliance, and format correctness. Evaluated against 17 baselines on a benchmark of 756 real-world functions, SmartCoder-R1 establishes a new state of the art, achieving top performance across five key metrics: a ComPass of 87.70%, a VulRate of 8.60%, a SafeAval of 80.16%, a FuncRate of 53.84%, and a FullRate of 50.53%. This FullRate marks a 45.79% relative improvement over the strongest baseline, DeepSeek-R1. Crucially, its generated reasoning also excels in human evaluations, achieving high-quality ratings for Functionality (82.7%), Security (85.3%), and Clarity (90.7%).

Autonomous vehicles must react in milliseconds while reasoning about road geometry and traffic intent to navigate complex situations. We introduce NovaDrive, a single-branch vision-language architecture that processes front-camera images, HD-map tiles, LiDAR depth, and textual waypoints in a single branch. A lightweight, two-stage cross-attention block first aligns waypoint tokens with the HD map, then refines attention over fine-grained image and depth patches. Coupled with a novel smoothness loss that discourages abrupt steering and speed changes, this design eliminates the need for recurrent memory. We fine-tune the top 15 layers of an 11B LLaMA-3.2 vision-language backbone, enabling real-time inference. On the nuScenes / Waymo subset of the MD-NEX Outdoor benchmark, NovaDrive raises success rate to 84% (+4%), boosts path-efficiency (SPL) to 0.66 (+0.11), and reduces collision frequency from 2.6% to 1.2% (-1.4%) relative to the previous state-of-the-art. Our ablations confirm that waypoint tokens, partial VLM fine-tuning, and the cross-attention fusion each contribute the most to these gains. Beyond safety, NovaDrive's shorter routes (resulting from the novel smoothness loss) translate to lower fuel or battery usage, pointing toward leaner, more easily updated driving stacks. NovaDrive can be extended to other embodied-AI domains as well.

Large language models with Mixture-of-Experts (MoE) architectures achieve efficiency and scalability, yet their routing mechanisms introduce safety alignment challenges insufficiently addressed by techniques developed for dense models. In this work, the MoE-specific safety risk of positional vulnerability-that safety-aligned behaviors rely on specific expert modules-is formalized and systematically analyzed. An analytical framework, SAFEx, is presented to robustly identify, characterize, and validate safety-critical experts via a stability-based expert selection procedure, and to decompose them into two functional groups: the Harmful Content Detection Group (HCDG), which specializes in identifying and recognizing harmful content within user inputs, and the Harmful Response Control Group (HRCG), which specializes in controlling and enforcing model behaviors to generate appropriate safety responses. Expert-level interventions are conducted to probe causality and to test mitigation. Targeted masking of SAFEx-selected experts reveals that safety behavior is highly concentrated. On Qwen3-30B-A3B, configured with 48 MoE-FFN layers and 128 experts per layer under top-8 routing (48x128=6,144 experts in total), disabling 12 selected experts reduces the refusal rate by 22%. In addition, lightweight adaptation is performed using LoRA under three configurations-the HRCG, the union of HCDG and HRCG, and all experts-and the resulting updates are composed through negative weight merging targeted at the HRCG, leading to improved refusal under adversarial prompts without full-model retraining. These results establish positional vulnerability as a distinct MoE-specific safety challenge and provide a practical, compute-efficient pathway for expert-level safety interventions within routed architectures.

Recently, major AI providers such as Google and OpenAI have introduced Finetuning-as-a-Service (FaaS), which allows users to customize Large Language Models (LLMs) using their own data. However, this service is vulnerable to safety degradation when user data includes harmful prompts, a threat known as harmful finetuning attacks. Prior works attempt to mitigate this issue by first constructing safety-aligned model and then finetuning the model on user data. However, we observe that the safety-aligned weights provide weak initialization for downstream task learning, leading to suboptimal safety-alignment and downstream task performance. To address this, we propose a Refusal-T eacher (Ref-T eacher)-guided finetuning framework. Instead of finetuning a safety-aligned model on user data, our approach directly finetunes the base model under the guidance of a safety-aligned Ref-Teacher, which filters harmful prompts from user data and distills safety-alignment knowledge into the base model. Extensive experiments demonstrate that our Ref-Teacher-guided finetuning strategy effectively minimizes harmful outputs and enhances finetuning accuracy for user-specific tasks, offering a practical solution for secure and reliable deployment of LLMs in FaaS. Recent advancements in Large Language Models (LLMs) (Touvron et al. (2023); Jiang et al. (2023); Team et al. (2024); Team (2024); Hurst et al. (2024); Guo et al. (2025); Research et al. (2025)) have achieved remarkable performance across a wide range of natural language processing tasks. LLMs are typically pretrained on massive and diverse corpora, resulting in strong generalization ability and broad applicability across domains. To further facilitate LLMs for individual and domain-specific purposes, major AI service providers such as Google and OpenAI offer not only access to pretrained LLMs but also Finetuning-as-a-Service (FaaS). This service enables users to upload custom datasets and adapt LLMs to more specific tasks and domains depending on their unique requirements. However, FaaS must prevent the malicious use of LLMs through safety-alignment, even when users attempt to jailbreak the models via customization. These types of attacks, which inject harmful prompts into user data for finetuning, are called harmful finetuning attacks. Several studies (Qi et al. (2023); Lermen et al. (2023); Rosati et al. (2024); Huang et al. (2024b;c;d); Li et al. (2025); Huang et al. (2025)) have shown that finetuning on user data containing harmful content compromises the safety-alignment, despite the LLMs being safety-aligned before finetuning.

Although Large Language Models (LLMs) perform well in general fields, they exhibit a confidence distortion problem on multi-choice question-answering (MCQA), particularly as the number of answer choices increases. Specifically, on MCQA with many choices, LLMs suffer from under-confidence in correct predictions and over-confidence in incorrect ones, leading to a substantially degraded performance. To solve this problem, we propose Self-ensemble in this work. Our method splits the choices into several groups and ensembles LLM predictions across these groups to reach a final decision. The advantage of Self-ensemble is its plug-and-play nature, where it can be integrated into existing LLM architecture based on a designed attention mask and positional encoding, without requiring labeled datasets for parameter tuning. Experimental results on three LLMs and datasets demonstrate that Self-ensemble comprehensively addresses the confidence distortion problem of LLMs, outperforming standard inference as well as baseline methods.

We present MA-RAG, a Multi-Agent framework for Retrieval-Augmented Generation (RAG) that addresses the inherent ambiguities and reasoning challenges in complex information-seeking tasks. Unlike conventional RAG methods that rely on end-to-end fine-tuning or isolated component enhancements, MA-RAG orchestrates a collaborative set of specialized AI agents: Planner, Step Definer, Extractor, and QA Agents, each responsible for a distinct stage of the RAG pipeline. By decomposing tasks into subtasks such as query disambiguation, evidence extraction, and answer synthesis, and enabling agents to communicate intermediate reasoning via chain-of-thought prompting, MA-RAG progressively refines retrieval and synthesis while maintaining modular interpretability. Extensive experiments on multi-hop and ambiguous QA benchmarks, including NQ, HotpotQA, 2WikimQA, and TriviaQA, demonstrate that MA-RAG significantly outperforms standalone LLMs and existing RAG methods across all model scales. Notably, even a small LLaMA3-8B model equipped with MA-RAG surpasses larger standalone LLMs, while larger variants (LLaMA3-70B and GPT-4o-mini) set new state-of-the-art results on challenging multi-hop datasets. Ablation studies reveal that both the planner and extractor agents are critical for multi-hop reasoning, and that high-capacity models are especially important for the QA agent to synthesize answers effectively. Beyond general-domain QA, MA-RAG generalizes to specialized domains such as medical QA, achieving competitive performance against domain-specific models without any domain-specific fine-tuning. Our results highlight the effectiveness of collaborative, modular reasoning in retrieval-augmented systems: MA-RAG not only improves answer accuracy and robustness but also provides interpretable intermediate reasoning steps, establishing a new paradigm for efficient and reliable multi-agent RAG.

Retrieval-Augmented Generation (RAG) leverages large language models (LLMs) combined with external contexts to enhance the accuracy and reliability of generated responses. However, reliably attributing generated content to specific context segments, context attribution, remains challenging due to the computationally intensive nature of current methods, which often require extensive fine-tuning or human annotation. In this work, we introduce a novel Jensen-Shannon Divergence driven method to Attribute Response to Context (ARC-JSD), enabling efficient and accurate identification of essential context sentences without additional fine-tuning, gradient-calculation or surrogate modelling. Evaluations on a wide range of RAG benchmarks, such as TyDi QA, Hotpot QA, and Musique, using instruction-tuned LLMs in different scales demonstrate superior accuracy and significant computational efficiency improvements compared to the previous surrogate-based method. Furthermore, our mechanistic analysis reveals specific attention heads and multilayer perceptron (MLP) layers responsible for context attribution, providing valuable insights into the internal workings of RAG models and how they affect RAG behaviours. Our code is available at https://github.com/ruizheliUOA/ARC_JSD.

Oversampling is one of the most widely used approaches for addressing imbalanced classification. The core idea is to generate additional minority samples to rebalance the dataset. Most existing methods, such as SMOTE, require converting categorical variables into numerical vectors, which often leads to information loss. Recently, large language model (LLM)-based methods have been introduced to overcome this limitation. However, current LLM-based approaches typically generate minority samples with limited diversity, reducing robustness and generalizability in downstream classification tasks. To address this gap, we propose a novel LLM-based oversampling method designed to enhance diversity. First, we introduce a sampling strategy that conditions synthetic sample generation on both minority labels and features. Second, we develop a new permutation strategy for fine-tuning pre-trained LLMs. Third, we fine-tune the LLM not only on minority samples but also on interpolated samples to further enrich variability. Extensive experiments on 10 tabular datasets demonstrate that our method significantly outperforms eight SOTA baselines. The generated synthetic samples are both realistic and diverse. Moreover, we provide theoretical analysis through an entropy-based perspective, proving that our method encourages diversity in the generated samples.