Search for: All records

Large Language Models (LLMs) have achieved remarkable success in natural language tasks, yet understanding their reasoning processes re- mains a significant challenge. We address this by introducing XplainLLM, a dataset accom- panying an explanation framework designed to enhance LLM transparency and reliability. Our dataset comprises 24,204 instances where each instance interprets the LLM’s reasoning behavior using knowledge graphs (KGs) and graph attention networks (GAT), and includes explanations of LLMs such as the decoder- only Llama-3 and the encoder-only RoBERTa. XplainLLM also features a framework for gener- ating grounded explanations and the debugger- scores for multidimensional quality analysis. Our explanations include why-choose and why- not-choose components, reason-elements, and debugger-scores that collectively illuminate the LLM’s reasoning behavior. Our evaluations demonstrate XplainLLM’s potential to reduce hallucinations and improve grounded explana- tion generation in LLMs. XplainLLM is a re- source for researchers and practitioners to build trust and verify the reliability of LLM outputs. Our code and dataset are publicly available. 

Boron carbide (B4C) has been well studied both theoretically and experimentally in its bulk form due to its exceptional hardness and use as a high-temperature thermoelectric. However, the properties of its two-dimensional nanosheets are not well established. In this paper, using van der Waals-corrected density-functional theory simulations, we show that bulk B4C can be cleaved along different directions to form B4C nanosheets with low formation energies. We find that there is minimal dependence of formation energies on cleavage planes and surface terminations, even though the bulk is not van der Waals layered. This anomalous stability of B4C nanosheets is found to be a result of surface reconstructions that are unique to B-rich systems. While the density of states of the bulk B4C indicate that it is a semiconductor, the B4C nanosheets are found to be predominantly metallic. We attribute this metallic behavior to a redistribution of charges on the surface bonds of the films. The Seebeck coefficients of the B4C films remain comparable to those of the bulk and are nearly constant as a function of temperature. Our results provide guidance for experimental synthesis efforts and future application of B4C nanosheets in nanoelectronic and thermoelectric applications. 

Abstract Due to its peculiar and highly variable nature, the blazar 3C 454.3 has been extensively monitored by the WEBT team. Here, we present for the first time these long-term optical flux and color variability results using data acquired inB,V,R, andIbands over a time span of about two decades. We include data from WEBT collaborators and public archives such as SMARTS, Steward Observatory, and Zwicky Transient Facility. The data are binned and segmented to study the source over this long term when more regular sampling was available. During our study, the long-term spectral variability reveals a redder-when-brighter trend, which, however, stabilizes at a particular brightness cutoff of ∼14.5 mag in theIband, after which it saturates and evolves into a complex state. This trend indicates increasing dominance of jet emission over accretion disk (AD) emission until jet emission completely dominates. Plots of the variation in spectral index (followingF_ν∝ν^−α) reveal a bimodal distribution using a one-day binning. These correlate with two extreme phases of 3C 454.3, an outburst or high-flux state and a quiescent or low-flux state, which are respectively jet- and AD-dominated. We have also conducted intraday variability studies of nine light curves and found that six of them are variable. Discrete correlation function analysis between different pairs of optical wave bands peaks at zero lags, indicating cospatial emission in different optical bands.