Vision transformers have recently gained great success on various computer vision tasks; nevertheless, their high model complexity makes it challenging to deploy on resource-constrained devices. Quantization is an effective approach to reduce model complexity, and data-free quantization, which can address data privacy and security concerns during model deployment, has received widespread interest. Unfortunately, all existing methods, such as BN regularization, were designed for convolutional neural networks and cannot be applied to vision transformers with significantly different model architectures. In this paper, we propose PSAQ-ViT, a Patch Similarity Aware data-free Quantization framework for Vision Transformers, to enable the generation of "realistic" samples based on the vision transformer's unique properties for calibrating the quantization parameters. Specifically, we analyze the selfattention module's properties and reveal a general difference (patch similarity) in its processing of Gaussian noise and real images. The above insights guide us to design a relative value metric to optimize the Gaussian noise to approximate the real images, which are then utilized to calibrate the quantization parameters. Extensive experiments and ablation studies are conducted on various benchmarks to validate the effectiveness of PSAQ-ViT, which can even outperform the real-data-driven methods.
Background
Coprophagy plays a vital role in maintaining growth and development in many small herbivores. Here, we constructed a coprophagy model by dividing rabbits into three groups, namely, control group (CON), sham-coprophagy prevention group (SCP), and coprophagy prevention group (CP), to explore the effects of coprophagy prevention on growth performance and cecal microecology in rabbits.
Results
Results showed that CP treatment decreased the feed utilization and growth performance of rabbits. Serum total cholesterol and total triglyceride in the CP group were remarkably lower than those in the other two groups. Furthermore, CP treatment destroyed cecum villi and reduced the content of short-chain fatty acids (SCFAs) in cecum contents. Gut microbiota profiling showed significant differences in the phylum and genus composition of cecal microorganisms among the three groups. At the genus level, the abundance of Oscillospira and Ruminococcus decreased significantly in the CP group. Enrichment analysis of metabolic pathways showed a significantly up-regulated differential metabolic pathway (PWY-7315, dTDP-N-acetylthomosamine biosynthesis) in the CP group compared with that in the CON group. Correlation analysis showed that the serum biochemical parameters were positively correlated with the abundance of Oscillospira, Sutterella, and Butyricimonas but negatively correlated with the abundance of Oxalobacte and Desulfovibrio. Meanwhile, the abundance of Butyricimonas and Parabacteroidesde was positively correlated with the concentration of butyric acid in the cecum.
Conclusions
In summary, coprophagy prevention had negative effects on serum biochemistry and gut microbiota, ultimately decreasing the growth performance of rabbits. The findings provide evidence for further revealing the biological significance of coprophagy in small herbivorous mammals.
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.