GVD Corporation (United States)

Large Language Models (LLMs) have shown remarkable performance across various tasks, yet significant disparities remain for non-English languages, and especially native African languages. This paper addresses these disparities by creating approximately 1 million human-translated words of new benchmark data in 8 low-resource African languages, covering a population of over 160 million speakers of: Amharic, Bambara, Igbo, Sepedi (Northern Sotho), Shona, Sesotho (Southern Sotho), Setswana, and Tsonga. Our benchmarks are translations of Winogrande and three sections of MMLU: college medicine, clinical knowledge, and virology. Using the translated benchmarks, we report previously unknown performance gaps between state-of-the-art (SOTA) LLMs in English and African languages. Finally, using results from over 400 fine-tuned models, we explore several methods to reduce the LLM performance gap, including high-quality dataset fine-tuning (using an LLM-as-an-Annotator), cross-lingual transfer, and cultural appropriateness adjustments. Key findings include average mono-lingual improvements of 5.6% with fine-tuning (with 5.4% average mono-lingual improvements when using high-quality data over low-quality data), 2.9% average gains from cross-lingual transfer, and a 3.0% out-of-the-box performance boost on culturally appropriate questions. The publicly available benchmarks, translations, and code from this study support further research and development aimed at creating more inclusive and effective language technologies.

IntroductionElafibranor, a dual peroxisome proliferator-activated receptor (PPAR) alpha/delta agonist, is under investigation as a treatment for primary biliary cholangitis (PBC). In vitro studies demonstrated that elafibranor and its major active metabolite GFT1007 inhibit organic anion transporter polypeptide (OATP) 1B3. A phase I trial was conducted to evaluate the drug-drug interaction (DDI) potential of elafibranor as a perpetrator with atorvastatin, an OATP1B3, OATP1B1 and cytochrome P4503A (CYP3A) clinical substrate.MethodsIn this phase I, randomised, open-label trial (EudraCT: 2015–000723-88), healthy participants received single doses of atorvastatin 40 mg (Days 1 and 17) and once-daily doses of elafibranor 180 mg (Days 4–19) under fasting conditions. Pharmacokinetic (PK) parameters derived for each treatment included observed maximum plasma concentration (Cmax), area under the plasma concentration curve from administration up to the last quantifiable concentration at 72 hours (AUC0–72 h), and time to reach Cmax (tmax). All adverse events (AEs) were recorded.ResultsOf 25 white male participants, 24 completed the trial; median (range) age was 32.0 (19–45) years. There was no increase in the overall exposure of atorvastatin. Compared with atorvastatin administration alone, Cmax was 28% lower following co-administration with elafibranor, with a geometric mean ratio (GMR) of 72 (90% confidence interval [CI]: 60.1–87.2); mean (standard deviation [SD]): 19.7 (11.0) ng/mL vs 14.8 (11.4) ng/mL. Median (range) tmax increased from 0.67 (0.50–3.00) hours (h) to 1.00 (0.67–4.50) h. There was a small decrease of mean (SD) AUC0–72 h (12%) following co-administration, and the 90% CIs indicated similarity between the regimens: 81.3 (42.1) ng/mL*h vs 75.0 (51.5) ng/mL*h; GMR: 88 (90% CI: 82.2–93.3). Observed differences between the two regimens were not deemed clinically significant. A total of 16 non-serious and mild to moderate treatment-emergent AEs were reported by 11 participants; all were resolved without treatment discontinuation.ConclusionsCo-administration of atorvastatin and elafibranor 180 mg had minimal impact on the overall exposure of atorvastatin. Therefore, co-administration of elafibranor 80 mg (clinical dose) with OATP1B1, OATP1B3 and CYP3A substrates, such as other statins, is unlikely to cause clinically significant PK DDIs. Furthermore, co-administration of atorvastatin and elafibranor was well tolerated.

The World Health Organization (WHO) and US Centers for Disease Control and Prevention (CDC) recommend cleaning soiled surfaces with soap and water, followed by use of approved disinfectant. However, data are lacking on the potential efficacy of soapy water alone as a disinfectant for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which is relevant to low-resource settings where soapy water is prevalent for handwashing. To our knowledge, no appropriate biosafety level 1 (BSL-1) surrogate has been identified and confirmed for use in studies with soapy water and the highly infectious SARS-CoV-2. Therefore, our objectives were to determine the efficacy of soapy water alone as a disinfectant against SARS-CoV-2 and if commonly used BSL-1 bacteriophage surrogates could serve as a surrogate model for testing soapy water as a disinfectant. Overall, results indicate that SARS-CoV-2 can be reduced >4 log10 in suspension but only 0.50 log10 on a nonporous surface with 10 min of exposure to 0.5% soapy water. This difference is potentially attributed to less area of exposure on surfaces than in suspension. Phi6 (a verified SARS-CoV-2 surrogate for other disinfectants) was not appropriate for SARS-CoV-2 disinfection with soapy water. Further research is needed to determine an appropriate surrogate for SARS-CoV-2 disinfection with soapy water as disinfection of MS2 was similar to SARS-CoV-2 on surfaces only. Our work highlights the importance of confirming surrogates for each disinfectant used. Based on our results, we do not recommend a change to the current WHO and CDC surface disinfection protocols that recommend using soapy water to preclean a surface before applying disinfectant.

The National Renewable Energy Laboratory is actively developing and testing Immersive Industrialized Construction Environments (IICE) for construction automation and worker-machine interaction to investigate possible solutions and increase workforce productivity. At full scope and matured functionality, IICE allows us to accelerate the development of and better explore industrialized construction approaches such as prefabrication. IICE also enables wider adoption of energy-efficient products and Industry 4.0 construction automation through worker-machine interaction pilots. Industry 4.0 and industrialized construction approaches can encourage workforce specialization in energy efficiency construction, address the lack of multi-skilled workers, and increase workforce productivity with construction automation. However, recent attempts to integrate these concepts with the industry have only been moderately successful. To address this, focusing the pedagogy on using a digital twin, its digital models, and virtual reality could make the experience of continuing education on construction automation more affordable, accessible, scalable, immersive, and safer, and could greatly improve the efficiency and robustness of the building and construction industry. IICE accurately represents the realities of construction uncertainties without having to create full scale physical prototypes of machines. In this paper, we address the following research question: How can a digital twin and its models in virtual reality enhance the learning experience and productivity of energy efficiency construction workers to gain the skills in operating Industry 4.0 components such as construction automation and handling energy-efficient products in industrialized construction factories and on-site? We introduce original research on developing IICE and present preliminary findings from time and motion pilot studies.

Abstract Supply chain under demand uncertainty has been a challenging problem due to increased competition and market volatility in modern markets. Flexibility in planning decisions makes modular manufacturing a promising way to address this problem. In this work, the problem of multiperiod process and supply chain network design is considered under demand uncertainty. A mixed integer two‐stage stochastic programming problem is formulated with integer variables indicating the process design and continuous variables to represent the material flow in the supply chain. The problem is solved using a rolling horizon approach. Benders decomposition is used to reduce the computational complexity of the optimization problem. To promote risk‐averse decisions, a downside risk measure is incorporated in the model. The results demonstrate the several advantages of modular designs in meeting product demands. A pareto‐optimal curve for minimizing the objectives of expected cost and downside risk is obtained.

<div class="section abstract"><div class="htmlview paragraph">The film thickness-roughness ratio Λ has been used since the mid-1960s as a simplified criterion for the lubrication conditions in rolling bearings. However, due to an assumed Gaussian distribution of the roughness height amplitudes of the functional surfaces of rolling bearings and other unsubstantiated assumptions regarding the mechanisms of contact interaction, concerns have arisen about the representability of Λ in terms of lubrication. In this study, a more objective Λ-type ratio that does not depend on the law of roughness height distribution and the peculiarities of contact mechanics is introduced and defined as Λ<i>z</i>. The relevant range of the new Λ ratio is lower than that of its conventional counterpart, and it overlaps with contemporary theoretical and experimental results. The study exhibits unique profilometric data of the functional surfaces of the roller bearings, produced by the major bearing manufacturers. An extensive and precise inspection of the surface texture of a representative number of these bearings revealed a substantial deviation in their surface roughness height from the Gaussian distribution (especially evidenced by the large scatter of their kurtosis and skewness values beyond a normality tolerance).</div></div>

Stereochemical configuration is determined by the relationship of atoms in three-dimensional space, yet remains most commonly represented in two-dimensional media such as printed publications or computer screens. Recommendations are provided for the display of three-dimensional stereochemical information in two-dimensional diagrams in ways that avoid ambiguity and are likely to be understood correctly by all viewers. Examples are provided for all types of stereochemical configuration, with explanation of which styles are preferred and which should be avoided. Principal recommendations include: