CVS Health Corporation (previously CVS Corporation and CVS Caremark Corporation) is an American healthcare company that owns CVS Pharmacy, a retail pharmacy chain; CVS Caremark, a pharmacy benefits manager; and Aetna, a health insurance provider, among many other brands. The company's headquarters is in Woonsocket, Rhode Island.
Hallucinations, defined as instances where Large Language Models (LLMs) generate false or misleading content, pose a significant challenge that impacts the safety and trust of downstream applications. We introduce UQLM, a Python package for LLM hallucination detection using state-of-the-art uncertainty quantification (UQ) techniques. This toolkit offers a suite of UQ-based scorers that compute response-level confidence scores ranging from 0 to 1. This library provides an off-the-shelf solution for UQ-based hallucination detection that can be easily integrated to enhance the reliability of LLM outputs.
Uncertainty quantification has emerged as an effective approach to closed-book hallucination detection for LLMs, but existing methods are largely designed for short-form outputs and do not generalize well to long-form generation. We introduce a taxonomy for fine-grained uncertainty quantification in long-form LLM outputs that distinguishes methods by design choices at three stages: response decomposition, unit-level scoring, and response-level aggregation. We formalize several families of consistency-based black-box scorers, providing generalizations and extensions of existing methods. We also introduce FactScore-STEM-Geo, a new 400-question long-form QA dataset spanning four categories across STEM and Geography. In our experiments across multiple LLMs and datasets, we find 1) claim-response entailment consistently performs better or on par with more complex claim-level scorers, 2) claim-level scoring generally yields better results than sentence-level scoring, and 3) uncertainty-aware decoding is highly effective for improving the factuality of long-form outputs. Our framework clarifies relationships between prior methods, enables apples-to-apples comparisons, and provides practical guidance for selecting components for fine-grained UQ.
OBJECTIVE:To evaluate febrile seizure risk following monovalent COVID-19 mRNA vaccination among children aged 2-5 years. METHODS:The primary analysis evaluated children who had a febrile seizure outcome in the 0-1 days following COVID-19 vaccination. A self-controlled case series analysis was performed in three commercial insurance databases to compare the risk of seizure in the risk interval (0-1 days) to a control interval (8-63 days). The exposure of interest was receipt of dose 1 and/or dose 2 of monovalent COVID-19 mRNA vaccinations. The primary outcome was febrile seizure (0-1 day risk interval). A conditional Poisson regression model was used to compare outcome rates in risk and control intervals and estimate incidence rate ratios (IRR) and 95% confidence intervals (CIs). Meta-analyses were used to pool results across databases. RESULTS:The primary meta-analysis found a statistically significant increased incidence of febrile seizure, in the 0-1 days following mRNA-1273 vaccination compared to the control interval (IRR: 2.52, 95% CI: 1.35 to 4.69, risk difference (RD)/100,000 doses = 3.22 (95% CI -0.31 to 6.75)). For the BNT162b2 vaccination, the IRR was elevated but not statistically significant (IRR: 1.41, 95% CI: 0.48 to 4.11, RD/100,000 doses = -0.25 (95% CI -2.75 to 2.24). CONCLUSIONS:Among children aged 2-5 years, the analysis showed a small elevated incidence rate ratio of febrile seizures in the 0-1 days following the mRNA-1273 vaccination.
The absence of standardized quality measures of critical staples foods such as rice, wheat, fruit, and vegetables poses a threat to the health of consumers and breaks the market. To overcome this obstacle, the research presents a quantum- enhanced intelligent grading system depending on harvest quality. This system detects flaws, contamination, and freshness and only quality produce passes through the packaging process and does not spoil other good quality produce. The system allows farmers to obtain higher prices of their products by making sure that it is only the high quality produce that passes through the packaging process therefore minimizing the losses that occur after harvest. It also eliminates instances of middlemen interfering with produce to cause unfair treatment to farmers and boost supply chain confidence. This scalable solution can change the food quality control in both urban and rural markets due to better food safety and increased transparency, which would eventually lead to healthier consumption and more efficient food supply system.