187037 research outputs found
Sort by
VISION-XL: High Definition Video Inverse Problem Solver using Latent Image Diffusion Models
No full textAdvancing Biochemical Engineering: Innovations in Nucleic Acid Engineering Over Three Decades
No full text
Single Ground Truth Is Not Enough: Add Linguistic Variability to Aspect-based Sentiment Analysis Evaluation
No full textOPT-OUT: Investigating Entity-Level Unlearning for Large Language Models via Optimal Transport
No full textInstruction-following large language models (LLMs), such as ChatGPT, have become widely popular among everyday users. However, these models inadvertently disclose private, sensitive information to their users, underscoring the need for machine unlearning techniques to remove selective information from the models. While prior work has focused on forgetting small, random subsets of training data at the instance-level, we argue that real-world scenarios often require the removal of an entire user data, which may require a more careful maneuver. In this study, we explore entity-level unlearning, which aims to erase all knowledge related to a target entity while preserving the remaining model capabilities. To address this, we introduce OPT-OUT, an optimal transport-based unlearning method that utilizes the Wasserstein distance from the model's initial parameters to achieve more effective and fine-grained unlearning. We also present the first Entity-Level Unlearning Dataset (ELUDe) designed to evaluate entity-level unlearning. Our empirical results demonstrate that OPT-OUT surpasses existing methods, establishing a new standard for secure and adaptable LLMs that can accommodate user data removal requests without the need for full retraining
Significance of transition metal dichalcogenides and their role in the activity of semiconductor materials for spectacular photocatalytic hydrogen production
No full textThe transition metal dichalcogenides (TMDs) are an important class of two-dimensional materials due to their tunable band gap, high carrier mobility, and adjustable carrier concentration. Owing to these advantages, TMD can be utilized in a wide range of applications. In this work, we discuss the role of TMDs and their modifications on semiconductor materials for photocatalytic activity studies. Several modification strategies of MoS2 have been explored to enhance its photocatalytic activity. These include: (i) deposition of few-layered MoS2 on CdS to increase surface-active site density (FMC), (ii) activating basal plane sites in addition to edge sites via Cu doping (Cu-FMC), and (iii) coupling with conductive reduced graphene oxide to facilitate charge transport while retaining catalytic edge sites (RGO-FMC). Comparative photocatalytic studies under solar light irradiation with lactic acid as a hole scavenger reveal that Cu doping enhances the intrinsic activity of MoS2, while reduced graphene oxide suppresses electron-hole recombination, and their combined structural modifications significantly boost hydrogen evolution performance, providing valuable insights into the rational design of transition metal sulfide-based heterostructures for solar fuel production.
Assessing the Impact of Climate Modes on Extreme Arctic Sea Ice Using Reanalysis Data
No full textArctic Sea ice variability arises from both anthropogenic forcing and natural climate modes such as the El Ni & ntilde;o-Southern Oscillation (ENSO), North Atlantic Oscillation (NAO) and Arctic Oscillation (AO). While these modes are known to influence sea ice concentration (SIC) and thickness (SIT), their impacts on seasonal extremes remain less understood. In this study, the extreme SIC and SIT are investigated using ERA5 and CMEMS reanalysis products, applying a non-stationary generalised extreme value (GEV) framework with climate indices as covariates. Results indicate that winter and spring sea-ice variability is most pronounced in the Barents and Greenland Seas, where strong Atlantic inflows and dynamic atmospheric conditions make the marginal ice zone highly sensitive to even minor perturbations. Conversely, in the central and peripheral Arctic, variability maximises in summer and autumn, when melt processes, ice-albedo feedback and delayed freeze-up intensify interannual fluctuations. ENSO exerts notable seasonal effects: El Ni & ntilde;o events enhance extreme SIC in the Laptev Sea but reduce it in the East Siberian Sea during summer, while SIT extremes increase in the Canadian Arctic Archipelago (CAA) and decline in the East Siberian Sea across all seasons. NAO-related anomalies include stronger SIC and SIT extremes in the Beaufort Sea and CAA and reductions in the Chukchi Sea during autumn. AO effects include increased SIC in the Chukchi Sea during summer and autumn, but decreases in the Beaufort and CAA in summer; SIT extremes rise in the CAA during spring but fall in the Beaufort Sea in summer. Composite analysis further reveals that out-of-phase NAO-AO states intensify autumn sea ice extremes, whereas in-phase conditions exert weaker influences. These results emphasise the distinct and seasonally varying roles of climate modes in shaping Arctic Sea ice extremes, offering insights into future Arctic climate variability.
Homogeneously Blended Donor and Acceptor AgBiS2 Nanocrystal Inks Enable High-Performance Eco-Friendly Solar Cells with Enhanced Carrier Diffusion Length
No full textColloidal semiconductor nanocrystals (NCs) have garnered significant attention as promising photovoltaic materials due to their tunable optoelectronic properties enabled by surface chemistry. Among them, AgBiS2 NCs stand out as an attractive candidate for solar cell applications due to their environmentally friendly composition, high absorption coefficients, and low-temperature processability. However, AgBiS2 NC photovoltaics generally exhibit lower power conversion efficiency (PCE) compared to other NC-based devices, primarily due to numerous surface traps that serve as recombination sites, leading to a short diffusion length for free carriers. To address this challenge, this work develops donor and acceptor blended (D/A) AgBiS2 films. Through ligand modulation, this work formulates acceptor and donor AgBiS2 NC inks with suitable electrical band alignment for charge separation, while ensuring that they are fully miscible in the same solvent. This enabled the fabrication of high-quality, thickness-controllable D/A-blended junction films. This work finds that this approach effectively facilitates carrier separation, leading to an enhanced carrier lifetime and diffusion length. As a result, using this approach, this work achieves AgBiS2 films that are twice as thick in solar cell applications compared to conventional devices, leading to improvements in current density and a solar cell PCE of 8.26%.
