Exploiting Syntactic Information to Boost the Fine-tuning of Pre-trained Models | IEEE Conference Publication | IEEE Xplore