We study generating abstractive summaries that are faithful and factually consistent with the given articles. A novel contrastive learning formulation is presented, which leverages both reference summaries, as positive training data, and automatically generated erroneous summaries, as negative training data, to train summarization systems that are better at distinguishing between them. We further design four types of strategies for creating negative samples, to resemble errors made commonly by two state-of-the-art models, BART and PEGASUS, found in our new human annotations of summary errors. Experiments on XSum and CNN/Daily Mail show that our contrastive learning framework is robust across datasets and models. It consistently produces more factual summaries than strong comparisons with post error correction, entailmentbased reranking, and unlikelihood training, according to QA-based factuality evaluation. Human judges echo the observation and find that our model summaries correct more errors.
Nanocarrier-based drug delivery systems are playing an emerging role in human immunodeficiency virus (HIV) chemoprophylaxis and treatment due to their ability to alter the pharmacokinetics and improve the therapeutic index of various antiretroviral (ARV) drug compounds used alone and in combination. Although several nanocarriers have been described for combination delivery of ARV drugs, measurement of drug-drug activities facilitated by the use of these nanotechnology platforms has not been fully investigated for topical prevention. Here, we show that physicochemically diverse ARV drugs can be encapsulated within polymeric nanoparticles to deliver multidrug combinations that provide potent HIV chemoprophylaxis in relevant models of cell-free, cell-cell, and mucosal tissue infection. In contrast to existing approaches that coformulate ARV drug combinations together in a single nanocarrier, we prepared single-drug-loaded nanoparticles that were subsequently combined upon administration. ARV drug-nanoparticles were prepared using emulsion-solvent evaporation techniques to incorporate maraviroc (MVC), etravirine (ETR), and raltegravir (RAL) into poly(lactic-co-glycolic acid) (PLGA) nanoparticles. We compared the antiviral potency of the free and formulated drug combinations for all pairwise and triple drug combinations against both cell-free and cell-associated HIV-1 infection in vitro. The efficacy of ARV-drug nanoparticle combinations was also assessed in a macaque cervicovaginal explant model using a chimeric simian-human immunodeficiency virus (SHIV) containing the reverse transcriptase (RT) of HIV-1. We observed that our ARV-NPs maintained potent HIV inhibition and were more effective when used in combinations. In particular, ARV-NP combinations involving ETR-NP exhibited significantly higher antiviral potency and dose-reduction against both cell-free and cell-associated HIV-1 BaL infection in vitro. Furthermore, ARV-NP combinations that showed large dose-reduction were identified to be synergistic, whereas the equivalent free-drug combinations were observed to be strictly additive. Higher intracellular drug concentration was measured for cells dosed with the triple ARV-NP combination compared to the equivalent unformulated drugs. Finally, as a first step toward evaluating challenge studies in animal models, we also show that our ARV-NP combinations inhibit RT-SHIV virus propagation in macaque cervicovaginal tissue and block virus transmission by migratory cells emigrating from the tissue. Our results demonstrate that ARV-NP combinations control HIV-1 transmission more efficiently than free-drug combinations. These studies provide a rationale to better understand the role of nanocarrier systems in facilitating multidrug effects in relevant cells and tissues associated with HIV infection.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.