Search

Long document summarization poses obstacles to current generative transformer-based models because of the broad context to process and understand. Indeed, detecting long-range dependencies is still challenging for today's state-of-the-art solutions, usually requiring model expansion at the cost of an unsustainable demand for computing and memory capacities. This paper introduces Emma, a novel efficient memory-enhanced transformer-based architecture. By segmenting a lengthy input into multiple text fragments, our model stores and compares the current chunk with previous ones, gaining the capability to read and comprehend the entire context over the whole document with a fixed amount of GPU memory. This method enables the model to deal with theoretically infinitely long documents, using less than 18 and 13 GB of memory for training and inference, respectively. We conducted extensive performance analyses and demonstrate that Emma achieved competitive results on two datasets of different domains while consuming significantly less GPU memory than competitors do, even in low-resource settings.

Positive gamma-aminobutyric acid type B (GABA B ) receptor modulators such as GS39783 have showed anxiolytic-like effects in several studies while such effects were absent in other studies. These conflicting findings led us hypothesize that the anxiolytic-like effects of such compounds depend on the individual basal anxiety and/or the anxiogenic properties of the used tests. The present study addresses this hypothesis by testing GS39783 effects on mice's anxiety-like behavior in a light-dark box. We found that GS39783 had no effects on a whole-group level. However, after grouping the mice for their basal anxiety, GS39783 reduced anxiety-like behavior in the subgroup with highest basal anxiety. Moreover, GS39783 effects correlated with individual basal anxiety. Next, the anxiogenic properties of the light-dark box test were increased by prior stress exposure. Again, GS39783 was not effective on a whole-group level. However, GS39783 had an anxiolytic-like effect in the most stress-responsive subgroup. Moreover, GS39783 effects correlated with individual stress responsiveness. Finally, we show that GS39783 brain levels were within a behaviorally relevant range. Overall, our study demonstrates that GS39783 effects depend on individual basal anxiety and stress responsiveness. This suggests that anxiety tests should generally be designed to capture individual basal anxiety and/or stress responsiveness as well as individual compound effects.

Perfluoroalkyl acids (PFAAs) are persistent man-made chemicals, ubiquitous in nature and present in human samples. Although restrictions are being introduced, they are still used in industrial processes as well as in consumer products. PFAAs cross the blood-brain-barrier and have been observed to induce adverse neurobehavioural effects in humans and animals as well as adverse effects in neuronal in vitro studies. The sulfonated PFAA perfluorooctane sulfonic acid (PFOS), has been shown to induce excitotoxicity via the N-methyl-D-aspartate receptor (NMDA-R) in cultures of rat cerebellar granule neurons (CGNs). In the present study the aim was to further characterise PFOS-induced toxicity (1-60 μM) in rat CGNs, by examining interactions between PFOS and elements of glutamatergic signalling and excitotoxicity. Effects of the carboxylated PFAA, perfluorooctanoic acid (PFOA, 300-500 μM) on the same endpoints were also examined. During experiments in immature cultures at days in vitro (DIV) 8, PFOS increased both the potency and efficacy of glutamate, whereas in mature cultures at DIV 14 only increased potency was observed. PFOA also increased potency at DIV 14. PFOS-enhanced glutamate toxicity was further antagonised by the competitive NMDA-R antagonist 3-((R)-2-Carboxypiperazin-4-yl)-propyl-1-phosphonic acid (CPP) at DIV 8. At DIV 8, PFOS also induced glutamate release (9-13 fold increase vs DMSO control) after 1-3 and 24 h exposure, whereas for PFOA a large (80 fold) increase was observed, but only after 24 h. PFOS and PFOA both also increased alanine and decreased serine levels after 24 h exposure. In conclusion, our results indicate that PFOS at concentrations relevant in an occupational setting, may be inducing excitotoxicity, and potentiation of glutamate signalling, via an allosteric action on the NMDA-R or by actions on other elements regulating glutamate release or NMDA-R function. Our results further support our previous findings that PFOS and PFOA at equipotent concentrations induce toxicity via different mechanisms of action., (Copyright © 2020 The Authors. Published by Elsevier B.V. All rights reserved.)