Search for: All records

The downscaling of silicon field-effect transistors is approaching fundamental limits, with the international roadmap for devices and systems (IRDS) projecting a 12 nm channel length plateau by 2037. Two-dimensional (2D) semiconductors have emerged as promising candidates for extended transistor scaling. However, no experimental demonstration has thus far surpassed the fundamental switching limits of advanced silicon. Here, we report monolayer MoS₂ transistors with physical channel lengths below 5 nm, achieving a subthreshold swing of 88 mV/dec and an on/off ratio above 106. These results surpass both the theoretical subthreshold limit of silicon and state-of-the-art emerging semiconductors-based transistors at comparable dimensions. To ensure high throughput, we develop a photolithography based, complementary metal oxide semiconductor (CMOS) compatible process and demonstrate 4-inch wafer-scale integration. We implement a design–technology co-optimization approach featuring an air-gap isolation and gate-contact overlap design to effectively suppress short channel effects. This work demonstrates subthreshold transport in a 2D semiconductor transistor beyond the silicon limit, showing a viable pathway for extending transistor scaling and energy-efficient computing. 

Resonant energy transfer (RET) between the equal parity 1s65s ${}^3{\mathrm{S}}_1$ and 1s66s ${}^3{\mathrm{S}}_1$ Rydberg levels in helium has been observed in low-temperature ( $\sim 80\mathrm{mK}$ ) collisions with ammonia molecules that undergo inversion transitions in their X ${}^1{\mathrm{A}}_1$ ground electronic state. This hybrid Rydberg-atom–polar-molecule resonant energy transfer represents a monopole-dipole energy exchange reaction that necessarily requires spatial overlap of the Rydberg-electron and molecular wave functions. Calculations that account explicitly for the charge-dipole interaction between the Rydberg electron and the molecule provide a quantitative explanation of the observations. Total parity is conserved in the reaction through the mixing of collisional angular momentum in the atom-molecule complex. This work opens opportunities to expand the toolbox for quantum science with charge-dipole-mediated energy exchange in hybrid atom–polar-molecule platforms. 

The propensity of large language models (LLMs) to generate hallucinations and non-factual content undermines their reliability in high-stakes domains, where rigorous control over Type I errors (the conditional probability of incorrectly classifying hallucinations as truthful content) is essential. Despite its importance, formal verification of LLM factuality with such guarantees remains largely unexplored. In this paper, we introduce FACTTEST, a novel framework that statistically assesses whether an LLM can provide correct answers to given questions with high-probability correctness guarantees. We formulate hallucina- tion detection as a hypothesis testing problem to enforce an upper bound of Type I errors at user-specified significance levels. Notably, we prove that FACTTEST also ensures strong Type II error control under mild conditions and can be extended to maintain its effectiveness when covariate shifts exist. FACTTEST is distribution-free and and model-agnostic. It works for any number of human-annotated samples and applies to any black-box or white-box LM. Extensive experiments demonstrate that FACTTEST effectively detects hallucinations and enable LLMs to abstain from answering unknown questions, leading to an over 40% accuracy improvement. 

Artificial Intelligence (AI) has demonstrated significant potential in healthcare, particularly in disease diagnosis and treatment planning. Recent progress in Medical Large Vision-Language Models (Med-LVLMs) has opened up new possibilities for interactive diagnostic tools. However, these models often suffer from factual hallucination, which can lead to incorrect diagnoses. Fine-tuning and retrieval-augmented generation (RAG) have emerged as methods to address these issues. However, the amount of high-quality data and distribution shifts between training data and deployment data limit the application of fine-tuning methods. Although RAG is lightweight and effective, existing RAG-based approaches are not sufficiently general to different medical domains and can potentially cause misalignment issues, both between modalities and between the model and the ground truth. In this paper, we propose a versatile multimodal RAG system, MMed-RAG, designed to enhance the factuality of Med-LVLMs. Our approach introduces a domain-aware retrieval mechanism, an adaptive retrieved contexts selection, and a provable RAG-based preference fine-tuning strategy. These innovations make the RAG process sufficiently general and reliable, significantly improving alignment when introducing retrieved contexts. Experimental results across five medical datasets (involving radiology, ophthalmology, pathology) on medical VQA and report generation demonstrate that MMed-RAG can achieve an average improvement of 43.8% in factual accuracy in the factual accuracy of Med-LVLMs.