Yan Wang

This paper provides the first comprehensive systematization and large-scale empirical evaluation of existing LLM-based Automated Penetration Testing (AutoPT) frameworks, offering a structured taxonomy and unified benchmark for the field.

The paper introduces Thinking as Compression (TaC), a novel paradigm showing that the inherent reasoning process of a large language model can naturally compress long context inputs, outperforming dedicated compression methods.

The paper introduces Thinking as Compression (TaC), a novel paradigm showing that the inherent reasoning process of a large language model can naturally compress long context inputs, outperforming dedicated compression methods.

The paper proposes SafeDIG, a robust safety steering framework that adapts Diffusion Transformers for text-to-image generation by treating safety control as position-aware sparse feature transfer, ensuring reliable safety across different risk domains.

The paper introduces StreamSynth, a sequential setting for synthetic data generation, and proposes SynLearner, a framework that enables LLMs to improve synthesis performance by accumulating and transferring experience across a stream of tasks.

The paper proposes SafeDIG, a robust safety steering framework that adapts Diffusion Transformers for text-to-image generation by treating safety control as position-aware sparse feature transfer, ensuring reliable safety across different risk domains.

The paper introduces StreamSynth, a sequential setting for synthetic data generation, and proposes SynLearner, a framework that enables LLMs to improve synthesis performance by accumulating and transferring experience across a stream of tasks.

The paper introduces ConsisGuard, a framework that addresses the 'deliberation-to-enforcement gap' in LLM guardrails by ensuring that the reasoning process is faithfully and consistently translated into the final safety decision.

The paper introduces ConsisGuard, a framework that addresses the 'deliberation-to-enforcement gap' in LLM guardrails by ensuring that the reasoning process is faithfully and consistently translated into the final safety decision.

WaveFilter is a novel, training-free framework that uses wavelet transforms to efficiently filter critical tokens in the KV cache, significantly improving the long-context performance of Diffusion LLMs.

WaveFilter is a novel, training-free framework that uses wavelet transforms to efficiently filter critical tokens in the KV cache, significantly improving the long-context performance of Diffusion LLMs.

The paper introduces InsightVQA, a large-scale benchmark dataset designed for hierarchical visual question answering that assesses complex emotion understanding and cognitive reasoning beyond simple emotion recognition.

The paper introduces InsightVQA, a large-scale benchmark dataset designed for hierarchical visual question answering that assesses complex emotion understanding and cognitive reasoning beyond simple emotion recognition.

The paper proposes OneReason, a framework that enhances the reasoning capability of generative recommendation models by focusing on improving item perception and structuring user behavior into coherent latent interests.

The paper proposes OneReason, a framework that enhances the reasoning capability of generative recommendation models by focusing on improving item perception and structuring user behavior into coherent latent interests.

This paper develops a unified topological toolkit for comparing neural representations, addressing limitations of existing methods.

The paper proposes OneReason, a framework that enhances the reasoning capability of generative recommendation models by focusing on improving item perception and structuring user behavior into coherent latent interests.

Libra is a load balancing approach for long-context LLM training that groups packed sequences into fixed-size pools and reduces attention workload variance, improving end-to-end throughput and straggler-attention speedup.

The paper introduces VeriFine, a framework that scales verification through the co-evolution of policy, training curriculum, and judge to support continuous self-improvement in embodied reasoning.