Class anchor clustering (CAC) is a loss that prompts the known training samples to form tight clusters around anchored, class-specific center points in the feature space.
SimCLR uses samples in the current mini-batch for negative example mining. The negative sample size is constrained with the minibatch size N.
Contrastive learning in the latent space has recently shown great promise, which aims to make the representation
Zero-shot stance detection (ZSSD) aims to detect stance for destination unseen targets by learning stance features from known targets
Stance detection aims to automatically identify one’s opinionated standpoint/attitude
Chen et al. (2018) proposed ADAN, an architecture based on a feed-forward neural network with three main components
Several works employed domain adaptation to improve performance.
Devlin et al. (2019) produced 2 BERT models, for English and Chinese. To support other languages, they trained a multilingual BERT (mBERT)
An integral part involved in developing various PLMs is providing NLU multitask benchmarks used to demonstrate the linguistic abilities of new models and approaches
Recent studies on the geometric properties of contextual embedding space have observed that the distribution of embedding vectors is far from isotropic and occupies a relatively narrow cone space