Gang Li

The paper introduces Cross-Model Neuron Transfer (CNT), a post-hoc method that efficiently transfers safety-oriented functionalities between different large language models by transferring minimal subsets of neurons, achieving high performance with minimal degradation.

The paper evaluates multi-LLM strategies for secure code generation, finding that hybrid pipelines combining ensembling, static analysis, and patching achieve the strongest security performance, outperforming single models and purely collaborative systems.

The paper proposes Trajectory Induced Preference Optimization (TIPO) to improve mobile GUI agent personalization by explicitly modeling and optimizing for privacy-related behavioral differences in execution trajectories.

The paper introduces a novel paradigm where a fine-tuned LLM acts as an ancillary predictor to forecast likely advertisers, significantly improving ad recommendation systems by augmenting candidate generation and providing priors for downstream ranking.

The paper introduces FinBoardBench, a novel evaluation suite using financial board games to demonstrate that current LLMs, despite strong static reasoning, fail at complex, dynamic wealth management and strategic decision-making.

The paper introduces CrystalXRD-Bench, a new benchmark designed to test Vision-Language Models (VLMs) on the complex task of identifying crystallographic Miller indices (HKLs) from rendered X-ray Diffraction (XRD) patterns, finding that current models struggle significantly with this multi-step scientific reasoning.

The paper proposes Agentic ASR, a closed-loop framework that treats ASR as a multi-turn refinement task, significantly improving semantic accuracy over traditional token-level metrics.

The paper introduces BilliardPhys-Bench, a new benchmark that demonstrates that current multimodal LLMs struggle with complex physical reasoning and predicting object dynamics in simulated environments.

The paper introduces Hierarchical Adaptive Budgeter (HAB), a framework that improves LLM reasoning efficiency by adaptively allocating computational resources to match the intrinsic complexity of both problems and individual reasoning steps.

The paper introduces AdvCL, a framework that repurposes adversarial perturbations as a geometric control signal to stabilize continual learning in large language models, significantly reducing forgetting and enhancing robustness.

IstGPT introduces a novel LLM-based framework for real-time, fine-grained anomaly detection in complex industrial cyber-physical systems, achieving state-of-the-art performance across multiple benchmarks.