11563 research outputs found
Sort by
Capturing the ring conformational dynamics of hexopyranoses: polarization effects, force field refinements and quantum mechanical investigations
No full text
Model hubs and beyond: analyzing model popularity, performance, and documentation
No full textWith the massive surge in ML models on platforms like Hugging Face, users often lose track and struggle to choose the best model for their downstream tasks, frequently relying on model popularity indicated by download counts, likes, or recency. We investigate whether this popularity aligns with actual model performance and how the comprehensiveness of model documentation correlates with both popularity and performance. In our study, we evaluated a comprehensive set of 500 Sentiment Analysis models on Hugging Face. This evaluation involved massive annotation efforts, with human annotators completing nearly 80,000 annotations, alongside extensive model training and evaluation. Our findings reveal that model popularity does not necessarily correlate with performance. Additionally, we identify critical inconsistencies in model card reporting: approximately 80\% of the models analyzed lack detailed information about the model, training, and evaluation processes. Furthermore, about 88\% of model authors overstate their models' performance in the model cards. Based on our findings, we provide a checklist of guidelines for users to choose good models for downstream tasks
Leaching Characteristics of Lime-Treated Fly Ash
No full textFly ash has been used in many geotechnical applications, such as fill material in embankments and ash dyke raising. However, fly ash alone may not provide sufficient strength in every project, and its treatment is required. Fly ash has been treated with lime as it is a pozzolanic material. Fly ash, a waste material of coal burning might contain heavy metals such as lead (Pb) and chromium (Cr) as impurities. Leaching of such heavy metal ions in water results in the degradation of the quality of water. Adding lime to fly ash results in surface modification of fly ash particles. Hence, evaluating the leaching characteristics of lime-treated fly ash becomes necessary. In the present study, leaching characteristics of lime-treated fly ash were studied for lime content of 1, 2, 3, 4, and 5%. The shear strength of lime-treated fly ash was also evaluated by performing a series of unconfined compression (UC) tests. The toxicity characteristic leaching procedure (TCLP) were followed to obtain a leachate solution of lime-treated fly ash. The ICP-OES technique was used to determine the metal ion concentrations in the solutions prepared. It was observed that the concentration of Ca, Mg, and Cr ions increased in leachate with the addition of lime. The concentration of Pb and Zn decreased with the addition of lime
SemFaceEdit: Semantic Face Editing on Generative Radiance Manifolds
No full textDespite multiple view consistency offered by 3D-aware GAN techniques, the resulting images often lack the capacity for localized editing. In response, generative radiance manifolds emerge as an efficient approach for constrained point sampling within volumes, effectively reducing computational demands and enabling the learning of fine details. This work introduces SemFaceEdit, a novel method that streamlines the appearance and geometric editing process by generating semantic fields on generative radiance manifolds. Utilizing latent codes, our method effectively disentangles the geometry and appearance associated with different facial semantics within the generated image. In contrast to existing methods that can change the appearance of the entire radiance field, our method enables the precise editing of particular facial semantics while preserving the integrity of other regions. Our network comprises two key modules: the Geometry module, which generates semantic radiance and occupancy fields, and the Appearance module, which is responsible for predicting RGB radiance. We jointly train both modules in adversarial settings to learn semantic-aware geometry and appearance descriptors. The appearance descriptors are then conditioned on their respective semantic latent codes by the Appearance Module, facilitating disentanglement and enhanced control. Our experiments highlight SemFaceEdit’s superior performance in semantic field-based editing, particularly in achieving improved radiance field disentanglement
Transient energy growth and resolvent analyses of the axisymmetric boundary layer with suction and injection
No full textThe present study involves bi-global non-modal and resolvent analyses of the incompressible boundary layer developing over a thin, long, horizontal circular cylinder under the influence of vectored, non-uniform wall transpiration. Transient growth and resolvent analyses are employed to identify the optimal structures of pair of input-output modes and associated energy gains over a time and frequency domains, respectively, providing insights into the dynamics of steady flow. The non-modal and resolvent gains are computed for three distinct transpiration profiles across five different vectored angles, corresponding to three different intensities and Reynolds numbers. The non-modal and resolvent energy gains are found to be higher for injection and lower for suction due to the amplification and decay of instability modes, respectively. Additionally, energy gains are found higher and lower for wall-normal uniform injection and suction profiles, respectively, compared to non-uniform profiles with other vectored angles in both analyses. Two-dimensional spatial structures of output modes in non-modal analysis reveal that the wave packets of optimal disturbances shift upstream and closer to the wall, with smaller cellular vortices for suction. In contrast, for injection, disturbances shift downstream and farther from the cylinder wall, with larger cellular vortices. The spatial structures of forcing-response modes indicate that the harmonic oscillations are located upstream, while the response characteristics are located downstream in the case of injection. However, in the case of suction, it is not possible to amplify the response modes downstream through harmonic forcing of the upstream flow due to the damping characteristics
Characterising Dissolution Dynamics of Engineered Nanomaterials: Advances in Analytical Techniques and Safety-by-Design
No full textEngineered Nanomaterials (ENM) have rapidly emerged as vital components in modern technology, most notably as vehicles in vaccine delivery, which highlights their growing potential for interaction with biological and environmental systems. One critical property influencing ENM behavior is dissolution, the release of ions and molecules into surrounding media, which dictates their abundance, fate, and biological response. A decade ago, dissolution was recognised as pivotal in understanding ENM interactions with exposure media and assessing their potential toxicity. Since then, progress in this field has led to a deeper understanding of ENM surface chemistry and transformations, positioning dissolution as a key factor in achieving “Safety-by-Design” (SbD) for sustainable ENM applications. Early dissolution studies relied on batch and flow-through methods, such as dialysis, but recent advances have favored in situ techniques such as single-cell/single-particle inductively coupled plasma mass spectrometry (ICP-MS) and liquid-cell electron microscopy, enabling real-time dissolution measurements. Additionally, computational models can now predict ENM reactivity and stability, enhancing the understanding of dissolution behavior. This perspective critically examines these developments, highlighting computational approaches for their efficiency and scalability, and proposes a roadmap to integrate these insights with SbD goals for safer, sustainable nanotechnology applications
Robust and High-Performance 12-T Interlocked SRAM for In-Memory Computing
No full textIn this paper, we analyze the existing SRAM based In-Memory Computing(IMC) proposals and show through exhaustive simulations that they fail under process variations. 6-T SRAM, 8-T SRAM, and 10-T SRAM based IMC architectures suffer from compute-disturb (stored data flips during IMC), compute-failure (provides false computation results), and half-select failures, respectively. To circumvent these issues, we propose a novel 12-T Dual Port Dual Interlockedstorage Cell (DPDICE) SRAM. DPDICE SRAM based IMC architecture(DPDICE-IMC) can perform essential boolean functions successfully in a single cycle and can perform basic arithmetic operations such as add and multiply. The most striking feature is that DPDICE-IMC architecture can perform IMC on two datasets simultaneously, thus doubling the throughput. Cumulatively, the proposed DPDICE-IMC is 26.7%, 8�, and 28% better than 6-T SRAM, 8-T SRAM, and 10-T SRAM based IMC architectures, respectively. � 2020 Elsevier B.V., All rights reserved